Upload 913 files

.gitattributes

@@ -36,3 +36,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 app/.github/examples/snapshot.mp4 filter=lfs diff=lfs merge=lfs -text
 app/docs/finaloutput.png filter=lfs diff=lfs merge=lfs -text
 app/temp/6efe43a0630bd48ea2e9c9a638f368e81348db7b/image.png filter=lfs diff=lfs merge=lfs -text
+docs/finaloutput.png filter=lfs diff=lfs merge=lfs -text
+temp/6efe43a0630bd48ea2e9c9a638f368e81348db7b/image.png filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -1,7 +1,7 @@
 FROM python:3.11
-WORKDIR /code
+WORKDIR /usr/src/app
 RUN apt-get update && apt-get install -y libgl1-mesa-glx
 COPY requirements.txt ./
 RUN pip install --no-cache-dir -r requirements.txt
 COPY . .
-CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]
+CMD ["python", "run.py"]

LICENSE

@@ -0,0 +1,661 @@
+                    GNU AFFERO GENERAL PUBLIC LICENSE
+                       Version 3, 19 November 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The GNU Affero General Public License is a free, copyleft license for
+software and other kinds of works, specifically designed to ensure
+cooperation with the community in the case of network server software.
+
+  The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works.  By contrast,
+our General Public Licenses are intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+  Developers that use our General Public Licenses protect your rights
+with two steps: (1) assert copyright on the software, and (2) offer
+you this License which gives you legal permission to copy, distribute
+and/or modify the software.
+
+  A secondary benefit of defending all users' freedom is that
+improvements made in alternate versions of the program, if they
+receive widespread use, become available for other developers to
+incorporate.  Many developers of free software are heartened and
+encouraged by the resulting cooperation.  However, in the case of
+software used on network servers, this result may fail to come about.
+The GNU General Public License permits making a modified version and
+letting the public access it on a server without ever releasing its
+source code to the public.
+
+  The GNU Affero General Public License is designed specifically to
+ensure that, in such cases, the modified source code becomes available
+to the community.  It requires the operator of a network server to
+provide the source code of the modified version running there to the
+users of that server.  Therefore, public use of a modified version, on
+a publicly accessible server, gives the public access to the source
+code of the modified version.
+
+  An older license, called the Affero General Public License and
+published by Affero, was designed to accomplish similar goals.  This is
+a different license, not a version of the Affero GPL, but Affero has
+released a new version of the Affero GPL which permits relicensing under
+this license.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                       TERMS AND CONDITIONS
+
+  0. Definitions.
+
+  "This License" refers to version 3 of the GNU Affero General Public License.
+
+  "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+  "The Program" refers to any copyrightable work licensed under this
+License.  Each licensee is addressed as "you".  "Licensees" and
+"recipients" may be individuals or organizations.
+
+  To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy.  The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+  A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+  To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy.  Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+  To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies.  Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+  An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License.  If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+  1. Source Code.
+
+  The "source code" for a work means the preferred form of the work
+for making modifications to it.  "Object code" means any non-source
+form of a work.
+
+  A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+  The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form.  A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+  The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities.  However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work.  For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+  The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+  The Corresponding Source for a work in source code form is that
+same work.
+
+  2. Basic Permissions.
+
+  All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met.  This License explicitly affirms your unlimited
+permission to run the unmodified Program.  The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work.  This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+  You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force.  You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright.  Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+  Conveying under any other circumstances is permitted solely under
+the conditions stated below.  Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+  No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+  When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+  4. Conveying Verbatim Copies.
+
+  You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+  You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+  5. Conveying Modified Source Versions.
+
+  You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+    a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+
+    b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under section
+    7.  This requirement modifies the requirement in section 4 to
+    "keep intact all notices".
+
+    c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy.  This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged.  This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+
+    d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+  A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit.  Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+  6. Conveying Non-Source Forms.
+
+  You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+    a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+
+    b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the
+    Corresponding Source from a network server at no charge.
+
+    c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source.  This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+
+    d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge.  You need not require recipients to copy the
+    Corresponding Source along with the object code.  If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source.  Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+
+    e) Convey the object code using peer-to-peer transmission, provided
+    you inform other peers where the object code and Corresponding
+    Source of the work are being offered to the general public at no
+    charge under subsection 6d.
+
+  A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+  A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling.  In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage.  For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product.  A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+  "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source.  The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+  If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information.  But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+  The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed.  Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+  Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+  7. Additional Terms.
+
+  "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law.  If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+  When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it.  (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.)  You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+  Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+    a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+
+    b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+
+    c) Prohibiting misrepresentation of the origin of that material, or
+    requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+
+    d) Limiting the use for publicity purposes of names of licensors or
+    authors of the material; or
+
+    e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+
+    f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions of
+    it) with contractual assumptions of liability to the recipient, for
+    any liability that these contractual assumptions directly impose on
+    those licensors and authors.
+
+  All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10.  If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term.  If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+  If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+  Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+  8. Termination.
+
+  You may not propagate or modify a covered work except as expressly
+provided under this License.  Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+  However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+  Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+  Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License.  If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+  9. Acceptance Not Required for Having Copies.
+
+  You are not required to accept this License in order to receive or
+run a copy of the Program.  Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance.  However,
+nothing other than this License grants you permission to propagate or
+modify any covered work.  These actions infringe copyright if you do
+not accept this License.  Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+  10. Automatic Licensing of Downstream Recipients.
+
+  Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License.  You are not responsible
+for enforcing compliance by third parties with this License.
+
+  An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations.  If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+  You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License.  For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+  11. Patents.
+
+  A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based.  The
+work thus licensed is called the contributor's "contributor version".
+
+  A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version.  For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+  Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+  In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement).  To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+  If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients.  "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+  If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+  A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License.  You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+  Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+  12. No Surrender of Others' Freedom.
+
+  If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all.  For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+  13. Remote Network Interaction; Use with the GNU General Public License.
+
+  Notwithstanding any other provision of this License, if you modify the
+Program, your modified version must prominently offer all users
+interacting with it remotely through a computer network (if your version
+supports such interaction) an opportunity to receive the Corresponding
+Source of your version by providing access to the Corresponding Source
+from a network server at no charge, through some standard or customary
+means of facilitating copying of software.  This Corresponding Source
+shall include the Corresponding Source for any work covered by version 3
+of the GNU General Public License that is incorporated pursuant to the
+following paragraph.
+
+  Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU General Public License into a single
+combined work, and to convey the resulting work.  The terms of this
+License will continue to apply to the part which is the covered work,
+but the work with which it is combined will remain governed by version
+3 of the GNU General Public License.
+
+  14. Revised Versions of this License.
+
+  The Free Software Foundation may publish revised and/or new versions of
+the GNU Affero General Public License from time to time.  Such new versions
+will be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+  Each version is given a distinguishing version number.  If the
+Program specifies that a certain numbered version of the GNU Affero General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation.  If the Program does not specify a version number of the
+GNU Affero General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+  If the Program specifies that a proxy can decide which future
+versions of the GNU Affero General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+  Later license versions may give you additional or different
+permissions.  However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+  15. Disclaimer of Warranty.
+
+  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. Limitation of Liability.
+
+  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+  17. Interpretation of Sections 15 and 16.
+
+  If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software: you can redistribute it and/or modify
+    it under the terms of the GNU Affero General Public License as published
+    by the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU Affero General Public License for more details.
+
+    You should have received a copy of the GNU Affero General Public License
+    along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper mail.
+
+  If your software can interact with users remotely through a computer
+network, you should also make sure that it provides a way for users to
+get its source.  For example, if your program is a web application, its
+interface could display a "Source" link that leads users to an archive
+of the code.  There are many ways you could offer source, and different
+solutions will be better for different programs; see section 13 for the
+specific requirements.
+
+  You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU AGPL, see
+<https://www.gnu.org/licenses/>.

README.md

@@ -1,11 +1,9 @@
 ---
-title: Face Swap Docker
-emoji: 🚀
-colorFrom: gray
-colorTo: yellow
-sdk: docker
-pinned: false
-license: gpl-3.0
+title: face-swap
+app_file: run.py
+sdk: gradio
+sdk_version: 3.40.1
 ---
+# roop-unleashed
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+WIP Version of roop-unleashed using Gradio UI

chain_img_processor/__init__.py

@@ -0,0 +1,4 @@
+from .image import ChainImgProcessor, ChainImgPlugin, get_single_image_processor, version
+from .video import ChainVideoProcessor, get_single_video_processor
+from .batchimage import ChainBatchImageProcessor
+from .ffmpeg_writer import FFMPEG_VideoWriter

chain_img_processor/batchimage.py

@@ -0,0 +1,86 @@
+from typing import Any, List, Callable
+import psutil
+import os
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from queue import Queue
+from .image import ChainImgProcessor
+from tqdm import tqdm
+import cv2
+
+def create_queue(temp_frame_paths: List[str]) -> Queue[str]:
+    queue: Queue[str] = Queue()
+    for frame_path in temp_frame_paths:
+        queue.put(frame_path)
+    return queue
+
+
+def pick_queue(queue: Queue[str], queue_per_future: int) -> List[str]:
+    queues = []
+    for _ in range(queue_per_future):
+        if not queue.empty():
+            queues.append(queue.get())
+    return queues
+
+
+
+class ChainBatchImageProcessor(ChainImgProcessor):
+    chain = None
+    func_params_gen = None
+    num_threads = 1
+
+    def __init__(self):
+        ChainImgProcessor.__init__(self)
+
+
+    def init_with_plugins(self):
+        self.init_plugins(["core"])
+        self.display_init_info()
+
+        init_on_start_arr = self.init_on_start.split(",")
+        for proc_id in init_on_start_arr:
+            self.init_processor(proc_id)
+
+    def update_progress(self, progress: Any = None) -> None:
+        process = psutil.Process(os.getpid())
+        memory_usage = process.memory_info().rss / 1024 / 1024 / 1024
+        progress.set_postfix({
+            'memory_usage': '{:.2f}'.format(memory_usage).zfill(5) + 'GB',
+            'execution_threads': self.num_threads
+        })
+        progress.refresh()
+        progress.update(1)
+
+
+    def process_frames(self, source_files: List[str], target_files: List[str], current_files, update: Callable[[], None]) -> None:
+        for f in current_files:
+            temp_frame = cv2.imread(f)
+            if temp_frame is not None:
+                if self.func_params_gen:
+                    params = self.func_params_gen(None, temp_frame)
+                else:
+                    params = {}
+                resimg, _ = self.run_chain(temp_frame, params, self.chain)
+                if resimg is not None:
+                    i = source_files.index(f)
+                    cv2.imwrite(target_files[i], resimg)
+            if update:
+                update()
+
+
+    def run_batch_chain(self, source_files, target_files, threads:int = 1, chain = None, params_frame_gen_func = None):
+        self.chain = chain
+        self.func_params_gen = params_frame_gen_func
+        progress_bar_format = '{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}{postfix}]'
+        total = len(source_files)
+        self.num_threads = threads
+        with tqdm(total=total, desc='Processing', unit='frame', dynamic_ncols=True, bar_format=progress_bar_format) as progress:
+            with ThreadPoolExecutor(max_workers=threads) as executor:
+                futures = []
+                queue = create_queue(source_files)
+                queue_per_future = max(len(source_files) // threads, 1)
+                while not queue.empty():
+                    future = executor.submit(self.process_frames, source_files, target_files, pick_queue(queue, queue_per_future), lambda: self.update_progress(progress))
+                    futures.append(future)
+                for future in as_completed(futures):
+                    future.result()
+

chain_img_processor/ffmpeg_writer.py

@@ -0,0 +1,253 @@
+"""
+FFMPEG_Writer - write set of frames to video file
+
+original from
+https://github.com/Zulko/moviepy/blob/master/moviepy/video/io/ffmpeg_writer.py
+
+removed unnecessary dependencies
+
+The MIT License (MIT)
+
+Copyright (c) 2015 Zulko
+Copyright (c) 2023 Janvarev Vladislav
+"""
+
+import os
+import subprocess as sp
+
+PIPE = -1
+STDOUT = -2
+DEVNULL = -3
+
+FFMPEG_BINARY = "ffmpeg"
+
+class FFMPEG_VideoWriter:
+    """ A class for FFMPEG-based video writing.
+
+    A class to write videos using ffmpeg. ffmpeg will write in a large
+    choice of formats.
+
+    Parameters
+    -----------
+
+    filename
+      Any filename like 'video.mp4' etc. but if you want to avoid
+      complications it is recommended to use the generic extension
+      '.avi' for all your videos.
+
+    size
+      Size (width,height) of the output video in pixels.
+
+    fps
+      Frames per second in the output video file.
+
+    codec
+      FFMPEG codec. It seems that in terms of quality the hierarchy is
+      'rawvideo' = 'png' > 'mpeg4' > 'libx264'
+      'png' manages the same lossless quality as 'rawvideo' but yields
+      smaller files. Type ``ffmpeg -codecs`` in a terminal to get a list
+      of accepted codecs.
+
+      Note for default 'libx264': by default the pixel format yuv420p
+      is used. If the video dimensions are not both even (e.g. 720x405)
+      another pixel format is used, and this can cause problem in some
+      video readers.
+
+    audiofile
+      Optional: The name of an audio file that will be incorporated
+      to the video.
+
+    preset
+      Sets the time that FFMPEG will take to compress the video. The slower,
+      the better the compression rate. Possibilities are: ultrafast,superfast,
+      veryfast, faster, fast, medium (default), slow, slower, veryslow,
+      placebo.
+
+    bitrate
+      Only relevant for codecs which accept a bitrate. "5000k" offers
+      nice results in general.
+
+    """
+
+    def __init__(self, filename, size, fps, codec="libx265", crf=14, audiofile=None,
+                 preset="medium", bitrate=None,
+                 logfile=None, threads=None, ffmpeg_params=None):
+
+        if logfile is None:
+            logfile = sp.PIPE
+
+        self.filename = filename
+        self.codec = codec
+        self.ext = self.filename.split(".")[-1]
+        w = size[0] - 1 if size[0] % 2 != 0 else size[0]
+        h = size[1] - 1 if size[1] % 2 != 0 else size[1]
+
+
+        # order is important
+        cmd = [
+            FFMPEG_BINARY,
+            '-hide_banner',
+            '-hwaccel', 'auto',
+            '-y',
+            '-loglevel', 'error' if logfile == sp.PIPE else 'info',
+            '-f', 'rawvideo',
+            '-vcodec', 'rawvideo',
+            '-s', '%dx%d' % (size[0], size[1]),
+            #'-pix_fmt', 'rgba' if withmask else 'rgb24',
+            '-pix_fmt', 'bgr24',
+            '-r', str(fps),
+            '-an', '-i', '-' 
+        ]
+
+        if audiofile is not None:
+            cmd.extend([
+                '-i', audiofile,
+                '-acodec', 'copy'
+            ])
+
+        cmd.extend([
+            '-vcodec', codec,
+            '-crf', str(crf)
+            #'-preset', preset,
+        ])
+        if ffmpeg_params is not None:
+            cmd.extend(ffmpeg_params)
+        if bitrate is not None:
+            cmd.extend([
+                '-b', bitrate
+            ])
+
+        # scale to a resolution divisible by 2 if not even
+        cmd.extend(['-vf', f'scale={w}:{h}' if w != size[0] or h != size[1] else 'colorspace=bt709:iall=bt601-6-625:fast=1'])
+
+        if threads is not None:
+            cmd.extend(["-threads", str(threads)])
+
+        cmd.extend([
+            '-pix_fmt', 'yuv420p',
+
+        ])
+        cmd.extend([
+            filename
+        ])
+
+        test = str(cmd)
+        print(test)
+
+        popen_params = {"stdout": DEVNULL,
+                        "stderr": logfile,
+                        "stdin": sp.PIPE}
+
+        # This was added so that no extra unwanted window opens on windows
+        # when the child process is created
+        if os.name == "nt":
+            popen_params["creationflags"] = 0x08000000  # CREATE_NO_WINDOW
+        
+        self.proc = sp.Popen(cmd, **popen_params)
+
+
+    def write_frame(self, img_array):
+        """ Writes one frame in the file."""
+        try:
+            #if PY3:
+            self.proc.stdin.write(img_array.tobytes())
+            # else:
+            #    self.proc.stdin.write(img_array.tostring())
+        except IOError as err:
+            _, ffmpeg_error = self.proc.communicate()
+            error = (str(err) + ("\n\nMoviePy error: FFMPEG encountered "
+                                 "the following error while writing file %s:"
+                                 "\n\n %s" % (self.filename, str(ffmpeg_error))))
+
+            if b"Unknown encoder" in ffmpeg_error:
+
+                error = error+("\n\nThe video export "
+                  "failed because FFMPEG didn't find the specified "
+                  "codec for video encoding (%s). Please install "
+                  "this codec or change the codec when calling "
+                  "write_videofile. For instance:\n"
+                  "  >>> clip.write_videofile('myvid.webm', codec='libvpx')")%(self.codec)
+
+            elif b"incorrect codec parameters ?" in ffmpeg_error:
+
+                 error = error+("\n\nThe video export "
+                  "failed, possibly because the codec specified for "
+                  "the video (%s) is not compatible with the given "
+                  "extension (%s). Please specify a valid 'codec' "
+                  "argument in write_videofile. This would be 'libx264' "
+                  "or 'mpeg4' for mp4, 'libtheora' for ogv, 'libvpx for webm. "
+                  "Another possible reason is that the audio codec was not "
+                  "compatible with the video codec. For instance the video "
+                  "extensions 'ogv' and 'webm' only allow 'libvorbis' (default) as a"
+                  "video codec."
+                  )%(self.codec, self.ext)
+
+            elif  b"encoder setup failed" in ffmpeg_error:
+
+                error = error+("\n\nThe video export "
+                  "failed, possibly because the bitrate you specified "
+                  "was too high or too low for the video codec.")
+
+            elif b"Invalid encoder type" in ffmpeg_error:
+
+                error = error + ("\n\nThe video export failed because the codec "
+                  "or file extension you provided is not a video")
+
+
+            raise IOError(error)
+
+    def close(self):
+        if self.proc:
+            self.proc.stdin.close()
+            if self.proc.stderr is not None:
+                self.proc.stderr.close()
+            self.proc.wait()
+
+        self.proc = None
+
+    # Support the Context Manager protocol, to ensure that resources are cleaned up.
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self.close()
+
+
+
+def ffmpeg_write_image(filename, image, logfile=False):
+    """ Writes an image (HxWx3 or HxWx4 numpy array) to a file, using
+        ffmpeg. """
+
+    if image.dtype != 'uint8':
+          image = image.astype("uint8")
+
+    cmd = [ FFMPEG_BINARY, '-y',
+           '-s', "%dx%d"%(image.shape[:2][::-1]),
+           "-f", 'rawvideo',
+           '-pix_fmt', "rgba" if (image.shape[2] == 4) else "rgb24",
+           '-i','-', filename]
+
+    if logfile:
+        log_file = open(filename + ".log", 'w+')
+    else:
+        log_file = sp.PIPE
+
+    popen_params = {"stdout": DEVNULL,
+                    "stderr": log_file,
+                    "stdin": sp.PIPE}
+
+    if os.name == "nt":
+        popen_params["creationflags"] = 0x08000000
+
+    proc = sp.Popen(cmd, **popen_params)
+    out, err = proc.communicate(image.tostring())
+
+    if proc.returncode:
+        err = "\n".join(["[MoviePy] Running : %s\n" % cmd,
+                         "WARNING: this command returned an error:",
+                         err.decode('utf8')])
+        raise IOError(err)
+
+    del proc
+    

chain_img_processor/image.py

@@ -0,0 +1,176 @@
+from jaa import JaaCore
+from roop.utilities import get_device
+
+
+from typing import Any
+
+version = "4.0.0"
+
+class ChainImgProcessor(JaaCore):
+
+    def __init__(self):
+        JaaCore.__init__(self)
+
+        self.processors:dict = {
+        }
+
+        self.processors_objects:dict[str,list[ChainImgPlugin]] = {}
+
+        self.default_chain = ""
+        self.init_on_start = ""
+
+        self.inited_processors = []
+
+        self.is_demo_row_render = False
+
+    def process_plugin_manifest(self, modname, manifest):
+        # adding processors from plugin manifest
+        if "img_processor" in manifest:  # process commands
+            for cmd in manifest["img_processor"].keys():
+                self.processors[cmd] = manifest["img_processor"][cmd]
+
+        return manifest
+
+    def init_with_plugins(self):
+        self.init_plugins(["core"])
+        self.display_init_info()
+
+        #self.init_translator_engine(self.default_translator)
+        init_on_start_arr = self.init_on_start.split(",")
+        for proc_id in init_on_start_arr:
+            self.init_processor(proc_id)
+
+    def run_chain(self, img, params:dict[str,Any] = None, chain:str = None, thread_index:int = 0):
+        if chain is None:
+            chain = self.default_chain
+        if params is None:
+            params = {}
+        params["_thread_index"] = thread_index
+        chain_ar = chain.split(",")
+        # init all not inited processors first
+        for proc_id in chain_ar:
+            if proc_id != "":
+                if not proc_id in self.inited_processors:
+                    self.init_processor(proc_id)
+
+
+
+        # run processing
+        if self.is_demo_row_render:
+            import cv2
+            import numpy as np
+            height, width, channels = img.shape
+            img_blank = np.zeros((height+30, width*(1+len(chain_ar)), 3), dtype=np.uint8)
+            img_blank.fill(255)
+
+            y = 30
+            x = 0
+            img_blank[y:y + height, x:x + width] = img
+
+            # Set the font scale and thickness
+            font_scale = 1
+            thickness = 2
+
+            # Set the font face to a monospace font
+            font_face = cv2.FONT_HERSHEY_SIMPLEX
+
+            cv2.putText(img_blank, "original", (x+4, y-7), font_face, font_scale, (0, 0, 0), thickness)
+
+
+        i = 0
+        for proc_id in chain_ar:
+            i += 1
+            if proc_id != "":
+                #img = self.processors[proc_id][1](self, img, params) # params can be modified inside
+                y = 30
+                img = self.processors_objects[proc_id][thread_index].process(img,params)
+                if self.is_demo_row_render:
+                    x = width*i
+                    img_blank[y:y + height, x:x + width] = img
+                    cv2.putText(img_blank, proc_id, (x + 4, y - 7), font_face, font_scale, (0, 0, 0), thickness)
+
+        if self.is_demo_row_render:
+            return img_blank, params
+
+        return img, params
+
+    # ---------------- init translation stuff ----------------
+    def fill_processors_for_thread_chains(self, threads:int = 1, chain:str = None):
+        if chain is None:
+            chain = self.default_chain
+
+        chain_ar = chain.split(",")
+        # init all not initialized processors first
+        for processor_id in chain_ar:
+            if processor_id != "":
+                if self.processors_objects.get(processor_id) is None:
+                    self.processors_objects[processor_id] = []
+                while len(self.processors_objects[processor_id]) < threads:
+                    self.add_processor_to_list(processor_id)
+
+    def add_processor_to_list(self, processor_id: str):
+        obj = self.processors[processor_id](self)
+        obj.init_plugin()
+        if self.processors_objects.get(processor_id) is None:
+            self.processors_objects[processor_id] = []
+        self.processors_objects[processor_id].append(obj)
+    def init_processor(self, processor_id: str):
+        if processor_id == "": # blank line case
+            return
+
+        if processor_id in self.inited_processors:
+            return
+
+        try:
+            if self.verbose:
+                self.print_blue("TRY: init processor plugin '{0}'...".format(processor_id))
+            self.add_processor_to_list(processor_id)
+            self.inited_processors.append(processor_id)
+            if self.verbose:
+                self.print_blue("SUCCESS: '{0}' initialized!".format(processor_id))
+
+        except Exception as e:
+            self.print_error("Error init processor plugin {0}...".format(processor_id), e)
+
+    # ------------ formatting stuff -------------------
+    def display_init_info(self):
+        if self.verbose:
+            print("ChainImgProcessor v{0}:".format(version))
+            self.format_print_key_list("processors:", self.processors.keys())
+
+    def format_print_key_list(self, key:str, value:list):
+        print(key+": ".join(value))
+
+    def print_error(self,err_txt,e:Exception = None):
+        print(err_txt,"red")
+        # if e != None:
+        #     cprint(e,"red")
+        import traceback
+        traceback.print_exc()
+
+    def print_red(self,txt):
+        print(txt)
+
+    def print_blue(self, txt):
+        print(txt)
+
+class ChainImgPlugin:
+
+    device = 'cpu'
+
+    def __init__(self, core: ChainImgProcessor):
+        self.core = core
+        self.device = get_device()
+
+    def init_plugin(self): # here you can init something. Called once
+        pass
+    def process(self, img, params:dict): # process img. Called multiple
+        return img
+
+_img_processor:ChainImgProcessor = None
+def get_single_image_processor() -> ChainImgProcessor:
+    global _img_processor
+    if _img_processor is None:
+        _img_processor = ChainImgProcessor()
+        _img_processor.init_with_plugins()
+    return _img_processor

chain_img_processor/video.py

@@ -0,0 +1,132 @@
+import roop.globals
+
+from threading import Thread
+from chain_img_processor import ChainImgProcessor
+
+class ThreadWithReturnValue(Thread):
+
+    def __init__(self, group=None, target=None, name=None,
+                 args=(), kwargs={}, Verbose=None):
+        Thread.__init__(self, group, target, name, args, kwargs)
+        self._return = None
+
+    def run(self):
+        if self._target is not None:
+            self._return = self._target(*self._args,
+                                        **self._kwargs)
+
+    def join(self, *args):
+        Thread.join(self, *args)
+        return self._return
+
+
+# in beta
+class ChainVideoProcessor(ChainImgProcessor):
+    def __init__(self):
+        ChainImgProcessor.__init__(self)
+
+        self.video_save_codec = "libx264"
+        self.video_save_crf = 14
+
+    def init_with_plugins(self):
+        self.init_plugins(["core","core_video"])
+        self.display_init_info()
+
+        init_on_start_arr = self.init_on_start.split(",")
+        for proc_id in init_on_start_arr:
+            self.init_processor(proc_id)
+
+    def run_video_chain(self, source_video, target_video, fps, threads:int = 1, chain = None, params_frame_gen_func = None, video_audio = None):
+        import cv2
+        from tqdm import tqdm
+        from chain_img_processor.ffmpeg_writer import FFMPEG_VideoWriter # ffmpeg install needed
+
+        cap = cv2.VideoCapture(source_video)
+        # width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
+        # height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
+        frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
+
+        # first frame do manually - because upscale may happen, we need to estimate width/height
+        ret, frame = cap.read()
+        if params_frame_gen_func is not None:
+            params = params_frame_gen_func(self, frame)
+        else:
+            params = {}
+        params["original_frame"] = frame
+        frame_processed, params = self.run_chain(frame,params,chain)
+        height, width, channels = frame_processed.shape
+
+        self.fill_processors_for_thread_chains(threads,chain)
+        #print(self.processors_objects)
+        #import threading
+        #locks:list[threading.Lock] = []
+        locks: list[bool] = []
+        for i in range(threads):
+            #locks.append(threading.Lock())
+            locks.append(False)
+
+        temp = []
+        with FFMPEG_VideoWriter(target_video, (width, height), fps, codec=roop.globals.video_encoder, crf=roop.globals.video_quality, audiofile=video_audio) as output_video_ff:
+            with tqdm(total=frame_count, desc='Processing', unit="frame", dynamic_ncols=True,
+                      bar_format='{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}{postfix}]') as progress:
+
+                # do first frame
+                output_video_ff.write_frame(frame_processed)
+                progress.update(1) #
+                cnt_frames = 0
+
+                # do rest frames
+                while True:
+                    # getting frame
+                    ret, frame = cap.read()
+
+                    if not ret:
+                        break
+                    cnt_frames+=1
+                    thread_ind = cnt_frames % threads
+                    # we are having an array of length %gpu_threads%, running in parallel
+                    # so if array is equal or longer than gpu threads, waiting
+                    #while len(temp) >= threads:
+                    while locks[thread_ind]:
+                        #print('WAIT', thread_ind)
+                        # we are order dependent, so we are forced to wait for first element to finish. When finished removing thread from the list
+                        frame_processed, params = temp.pop(0).join()
+                        locks[params["_thread_index"]] = False
+                        #print('OFF',cnt_frames,locks[params["_thread_index"]],locks)
+                        # writing into output
+                        output_video_ff.write_frame(frame_processed)
+                        # updating the status
+                        progress.update(1)
+
+                    # calc params for frame
+                    if params_frame_gen_func is not None:
+                        params = params_frame_gen_func(self,frame)
+                    else:
+                        params = {}
+
+                    # adding new frame to the list and starting it
+                    locks[thread_ind] = True
+                    #print('ON', cnt_frames, thread_ind, locks)
+                    params["original_frame"] = frame
+                    temp.append(
+                        ThreadWithReturnValue(target=self.run_chain, args=(frame, params, chain, thread_ind)))
+                    temp[-1].start()
+
+                while len(temp) > 0:
+                    # we are order dependent, so we are forced to wait for first element to finish. When finished removing thread from the list
+                    frame_processed, params = temp.pop(0).join()
+                    locks[params["_thread_index"]] = False
+                    # writing into output
+                    output_video_ff.write_frame(frame_processed)
+
+                    progress.update(1)
+
+                #print("FINAL", locks)
+
+_video_processor:ChainVideoProcessor = None
+def get_single_video_processor() -> ChainVideoProcessor:
+    global _video_processor
+    if _video_processor is None:
+        _video_processor = ChainVideoProcessor()
+        _video_processor.init_with_plugins()
+    return _video_processor

clip/__init__.py

@@ -0,0 +1 @@
+from .clip import *

clip/bpe_simple_vocab_16e6.txt.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:924691ac288e54409236115652ad4aa250f48203de50a9e4722a6ecd48d6804a
+size 1356917

clip/clip.py

@@ -0,0 +1,245 @@
+import hashlib
+import os
+import urllib
+import warnings
+from typing import Any, Union, List
+from pkg_resources import packaging
+
+import torch
+from PIL import Image
+from torchvision.transforms import Compose, Resize, CenterCrop, ToTensor, Normalize
+from tqdm import tqdm
+
+from .model import build_model
+from .simple_tokenizer import SimpleTokenizer as _Tokenizer
+
+try:
+    from torchvision.transforms import InterpolationMode
+    BICUBIC = InterpolationMode.BICUBIC
+except ImportError:
+    BICUBIC = Image.BICUBIC
+
+
+if packaging.version.parse(torch.__version__) < packaging.version.parse("1.7.1"):
+    warnings.warn("PyTorch version 1.7.1 or higher is recommended")
+
+
+__all__ = ["available_models", "load", "tokenize"]
+_tokenizer = _Tokenizer()
+
+_MODELS = {
+    "RN50": "https://openaipublic.azureedge.net/clip/models/afeb0e10f9e5a86da6080e35cf09123aca3b358a0c3e3b6c78a7b63bc04b6762/RN50.pt",
+    "RN101": "https://openaipublic.azureedge.net/clip/models/8fa8567bab74a42d41c5915025a8e4538c3bdbe8804a470a72f30b0d94fab599/RN101.pt",
+    "RN50x4": "https://openaipublic.azureedge.net/clip/models/7e526bd135e493cef0776de27d5f42653e6b4c8bf9e0f653bb11773263205fdd/RN50x4.pt",
+    "RN50x16": "https://openaipublic.azureedge.net/clip/models/52378b407f34354e150460fe41077663dd5b39c54cd0bfd2b27167a4a06ec9aa/RN50x16.pt",
+    "RN50x64": "https://openaipublic.azureedge.net/clip/models/be1cfb55d75a9666199fb2206c106743da0f6468c9d327f3e0d0a543a9919d9c/RN50x64.pt",
+    "ViT-B/32": "https://openaipublic.azureedge.net/clip/models/40d365715913c9da98579312b702a82c18be219cc2a73407c4526f58eba950af/ViT-B-32.pt",
+    "ViT-B/16": "https://openaipublic.azureedge.net/clip/models/5806e77cd80f8b59890b7e101eabd078d9fb84e6937f9e85e4ecb61988df416f/ViT-B-16.pt",
+    "ViT-L/14": "https://openaipublic.azureedge.net/clip/models/b8cca3fd41ae0c99ba7e8951adf17d267cdb84cd88be6f7c2e0eca1737a03836/ViT-L-14.pt",
+    "ViT-L/14@336px": "https://openaipublic.azureedge.net/clip/models/3035c92b350959924f9f00213499208652fc7ea050643e8b385c2dac08641f02/ViT-L-14-336px.pt",
+}
+
+
+def _download(url: str, root: str):
+    os.makedirs(root, exist_ok=True)
+    filename = os.path.basename(url)
+
+    expected_sha256 = url.split("/")[-2]
+    download_target = os.path.join(root, filename)
+
+    if os.path.exists(download_target) and not os.path.isfile(download_target):
+        raise RuntimeError(f"{download_target} exists and is not a regular file")
+
+    if os.path.isfile(download_target):
+        if hashlib.sha256(open(download_target, "rb").read()).hexdigest() == expected_sha256:
+            return download_target
+        else:
+            warnings.warn(f"{download_target} exists, but the SHA256 checksum does not match; re-downloading the file")
+
+    with urllib.request.urlopen(url) as source, open(download_target, "wb") as output:
+        with tqdm(total=int(source.info().get("Content-Length")), ncols=80, unit='iB', unit_scale=True, unit_divisor=1024) as loop:
+            while True:
+                buffer = source.read(8192)
+                if not buffer:
+                    break
+
+                output.write(buffer)
+                loop.update(len(buffer))
+
+    if hashlib.sha256(open(download_target, "rb").read()).hexdigest() != expected_sha256:
+        raise RuntimeError("Model has been downloaded but the SHA256 checksum does not not match")
+
+    return download_target
+
+
+def _convert_image_to_rgb(image):
+    return image.convert("RGB")
+
+
+def _transform(n_px):
+    return Compose([
+        Resize(n_px, interpolation=BICUBIC),
+        CenterCrop(n_px),
+        _convert_image_to_rgb,
+        ToTensor(),
+        Normalize((0.48145466, 0.4578275, 0.40821073), (0.26862954, 0.26130258, 0.27577711)),
+    ])
+
+
+def available_models() -> List[str]:
+    """Returns the names of available CLIP models"""
+    return list(_MODELS.keys())
+
+
+def load(name: str, device: Union[str, torch.device] = "cuda" if torch.cuda.is_available() else "cpu", jit: bool = False, download_root: str = None):
+    """Load a CLIP model
+
+    Parameters
+    ----------
+    name : str
+        A model name listed by `clip.available_models()`, or the path to a model checkpoint containing the state_dict
+
+    device : Union[str, torch.device]
+        The device to put the loaded model
+
+    jit : bool
+        Whether to load the optimized JIT model or more hackable non-JIT model (default).
+
+    download_root: str
+        path to download the model files; by default, it uses "~/.cache/clip"
+
+    Returns
+    -------
+    model : torch.nn.Module
+        The CLIP model
+
+    preprocess : Callable[[PIL.Image], torch.Tensor]
+        A torchvision transform that converts a PIL image into a tensor that the returned model can take as its input
+    """
+    if name in _MODELS:
+        model_path = _download(_MODELS[name], download_root or os.path.expanduser("~/.cache/clip"))
+    elif os.path.isfile(name):
+        model_path = name
+    else:
+        raise RuntimeError(f"Model {name} not found; available models = {available_models()}")
+
+    with open(model_path, 'rb') as opened_file:
+        try:
+            # loading JIT archive
+            model = torch.jit.load(opened_file, map_location=device if jit else "cpu").eval()
+            state_dict = None
+        except RuntimeError:
+            # loading saved state dict
+            if jit:
+                warnings.warn(f"File {model_path} is not a JIT archive. Loading as a state dict instead")
+                jit = False
+            state_dict = torch.load(opened_file, map_location="cpu")
+
+    if not jit:
+        model = build_model(state_dict or model.state_dict()).to(device)
+        if str(device) == "cpu":
+            model.float()
+        return model, _transform(model.visual.input_resolution)
+
+    # patch the device names
+    device_holder = torch.jit.trace(lambda: torch.ones([]).to(torch.device(device)), example_inputs=[])
+    device_node = [n for n in device_holder.graph.findAllNodes("prim::Constant") if "Device" in repr(n)][-1]
+
+    def _node_get(node: torch._C.Node, key: str):
+        """Gets attributes of a node which is polymorphic over return type.
+        
+        From https://github.com/pytorch/pytorch/pull/82628
+        """
+        sel = node.kindOf(key)
+        return getattr(node, sel)(key)
+
+    def patch_device(module):
+        try:
+            graphs = [module.graph] if hasattr(module, "graph") else []
+        except RuntimeError:
+            graphs = []
+
+        if hasattr(module, "forward1"):
+            graphs.append(module.forward1.graph)
+
+        for graph in graphs:
+            for node in graph.findAllNodes("prim::Constant"):
+                if "value" in node.attributeNames() and str(_node_get(node, "value")).startswith("cuda"):
+                    node.copyAttributes(device_node)
+
+    model.apply(patch_device)
+    patch_device(model.encode_image)
+    patch_device(model.encode_text)
+
+    # patch dtype to float32 on CPU
+    if str(device) == "cpu":
+        float_holder = torch.jit.trace(lambda: torch.ones([]).float(), example_inputs=[])
+        float_input = list(float_holder.graph.findNode("aten::to").inputs())[1]
+        float_node = float_input.node()
+
+        def patch_float(module):
+            try:
+                graphs = [module.graph] if hasattr(module, "graph") else []
+            except RuntimeError:
+                graphs = []
+
+            if hasattr(module, "forward1"):
+                graphs.append(module.forward1.graph)
+
+            for graph in graphs:
+                for node in graph.findAllNodes("aten::to"):
+                    inputs = list(node.inputs())
+                    for i in [1, 2]:  # dtype can be the second or third argument to aten::to()
+                        if _node_get(inputs[i].node(), "value") == 5:
+                            inputs[i].node().copyAttributes(float_node)
+
+        model.apply(patch_float)
+        patch_float(model.encode_image)
+        patch_float(model.encode_text)
+
+        model.float()
+
+    return model, _transform(model.input_resolution.item())
+
+
+def tokenize(texts: Union[str, List[str]], context_length: int = 77, truncate: bool = False) -> Union[torch.IntTensor, torch.LongTensor]:
+    """
+    Returns the tokenized representation of given input string(s)
+
+    Parameters
+    ----------
+    texts : Union[str, List[str]]
+        An input string or a list of input strings to tokenize
+
+    context_length : int
+        The context length to use; all CLIP models use 77 as the context length
+
+    truncate: bool
+        Whether to truncate the text in case its encoding is longer than the context length
+
+    Returns
+    -------
+    A two-dimensional tensor containing the resulting tokens, shape = [number of input strings, context_length].
+    We return LongTensor when torch version is <1.8.0, since older index_select requires indices to be long.
+    """
+    if isinstance(texts, str):
+        texts = [texts]
+
+    sot_token = _tokenizer.encoder["<|startoftext|>"]
+    eot_token = _tokenizer.encoder["<|endoftext|>"]
+    all_tokens = [[sot_token] + _tokenizer.encode(text) + [eot_token] for text in texts]
+    if packaging.version.parse(torch.__version__) < packaging.version.parse("1.8.0"):
+        result = torch.zeros(len(all_tokens), context_length, dtype=torch.long)
+    else:
+        result = torch.zeros(len(all_tokens), context_length, dtype=torch.int)
+
+    for i, tokens in enumerate(all_tokens):
+        if len(tokens) > context_length:
+            if truncate:
+                tokens = tokens[:context_length]
+                tokens[-1] = eot_token
+            else:
+                raise RuntimeError(f"Input {texts[i]} is too long for context length {context_length}")
+        result[i, :len(tokens)] = torch.tensor(tokens)
+
+    return result

clip/clipseg.py

@@ -0,0 +1,538 @@
+import math
+from os.path import basename, dirname, join, isfile
+import torch
+from torch import nn
+from torch.nn import functional as nnf
+from torch.nn.modules.activation import ReLU
+
+
+def get_prompt_list(prompt):
+    if prompt == 'plain':
+        return ['{}']    
+    elif prompt == 'fixed':
+        return ['a photo of a {}.']
+    elif prompt == 'shuffle':
+        return ['a photo of a {}.', 'a photograph of a {}.', 'an image of a {}.', '{}.']
+    elif prompt == 'shuffle+':
+        return ['a photo of a {}.', 'a photograph of a {}.', 'an image of a {}.', '{}.',
+                            'a cropped photo of a {}.', 'a good photo of a {}.', 'a photo of one {}.',
+                            'a bad photo of a {}.', 'a photo of the {}.']
+    else:
+        raise ValueError('Invalid value for prompt')        
+
+
+def forward_multihead_attention(x, b, with_aff=False, attn_mask=None):
+    """ 
+    Simplified version of multihead attention (taken from torch source code but without tons of if clauses). 
+    The mlp and layer norm come from CLIP.
+    x: input.
+    b: multihead attention module. 
+    """
+
+    x_ = b.ln_1(x)
+    q, k, v = nnf.linear(x_, b.attn.in_proj_weight, b.attn.in_proj_bias).chunk(3, dim=-1)
+    tgt_len, bsz, embed_dim = q.size()
+
+    head_dim = embed_dim // b.attn.num_heads
+    scaling = float(head_dim) ** -0.5
+
+    q = q.contiguous().view(tgt_len, bsz * b.attn.num_heads, b.attn.head_dim).transpose(0, 1)
+    k = k.contiguous().view(-1, bsz * b.attn.num_heads, b.attn.head_dim).transpose(0, 1)
+    v = v.contiguous().view(-1, bsz * b.attn.num_heads, b.attn.head_dim).transpose(0, 1)
+
+    q = q * scaling
+
+    attn_output_weights = torch.bmm(q, k.transpose(1, 2)) #  n_heads * batch_size, tokens^2, tokens^2
+    if attn_mask is not None:
+
+
+        attn_mask_type, attn_mask = attn_mask
+        n_heads = attn_output_weights.size(0) // attn_mask.size(0)
+        attn_mask = attn_mask.repeat(n_heads, 1)
+        
+        if attn_mask_type == 'cls_token':
+            # the mask only affects similarities compared to the readout-token.
+            attn_output_weights[:, 0, 1:] = attn_output_weights[:, 0, 1:] * attn_mask[None,...]
+            # attn_output_weights[:, 0, 0] = 0*attn_output_weights[:, 0, 0]
+
+        if attn_mask_type == 'all':
+            # print(attn_output_weights.shape, attn_mask[:, None].shape)
+            attn_output_weights[:, 1:, 1:] = attn_output_weights[:, 1:, 1:] * attn_mask[:, None]
+        
+    
+    attn_output_weights = torch.softmax(attn_output_weights, dim=-1)
+
+    attn_output = torch.bmm(attn_output_weights, v)
+    attn_output = attn_output.transpose(0, 1).contiguous().view(tgt_len, bsz, embed_dim)
+    attn_output = b.attn.out_proj(attn_output)
+
+    x = x + attn_output
+    x = x + b.mlp(b.ln_2(x))
+
+    if with_aff:
+        return x, attn_output_weights
+    else:
+        return x
+
+
+class CLIPDenseBase(nn.Module):
+
+    def __init__(self, version, reduce_cond, reduce_dim, prompt, n_tokens):
+        super().__init__()
+
+        import clip
+
+        # prec = torch.FloatTensor
+        self.clip_model, _ = clip.load(version, device='cpu', jit=False)
+        self.model = self.clip_model.visual
+
+        # if not None, scale conv weights such that we obtain n_tokens.
+        self.n_tokens = n_tokens
+
+        for p in self.clip_model.parameters():
+            p.requires_grad_(False)
+
+        # conditional
+        if reduce_cond is not None:
+            self.reduce_cond = nn.Linear(512, reduce_cond)
+            for p in self.reduce_cond.parameters():
+                p.requires_grad_(False)
+        else:
+            self.reduce_cond = None        
+
+        self.film_mul = nn.Linear(512 if reduce_cond is None else reduce_cond, reduce_dim)
+        self.film_add = nn.Linear(512 if reduce_cond is None else reduce_cond, reduce_dim)
+        
+        self.reduce = nn.Linear(768, reduce_dim)
+
+        self.prompt_list = get_prompt_list(prompt)     
+
+        # precomputed prompts
+        import pickle
+        if isfile('precomputed_prompt_vectors.pickle'):
+            precomp = pickle.load(open('precomputed_prompt_vectors.pickle', 'rb'))
+            self.precomputed_prompts = {k: torch.from_numpy(v) for k, v in precomp.items()}        
+        else:
+            self.precomputed_prompts = dict()
+    
+    def rescaled_pos_emb(self, new_size):
+        assert len(new_size) == 2
+
+        a = self.model.positional_embedding[1:].T.view(1, 768, *self.token_shape)
+        b = nnf.interpolate(a, new_size, mode='bicubic', align_corners=False).squeeze(0).view(768, new_size[0]*new_size[1]).T
+        return torch.cat([self.model.positional_embedding[:1], b])
+
+    def visual_forward(self, x_inp, extract_layers=(), skip=False, mask=None):
+        
+
+        with torch.no_grad():
+
+            inp_size = x_inp.shape[2:]
+
+            if self.n_tokens is not None:
+                stride2 = x_inp.shape[2] // self.n_tokens
+                conv_weight2 = nnf.interpolate(self.model.conv1.weight, (stride2, stride2), mode='bilinear', align_corners=True)
+                x = nnf.conv2d(x_inp, conv_weight2, bias=self.model.conv1.bias, stride=stride2, dilation=self.model.conv1.dilation)
+            else:
+                x = self.model.conv1(x_inp)  # shape = [*, width, grid, grid]
+
+            x = x.reshape(x.shape[0], x.shape[1], -1)  # shape = [*, width, grid ** 2]
+            x = x.permute(0, 2, 1)  # shape = [*, grid ** 2, width]
+
+            x = torch.cat([self.model.class_embedding.to(x.dtype) + torch.zeros(x.shape[0], 1, x.shape[-1], dtype=x.dtype, device=x.device), x], dim=1)  # shape = [*, grid ** 2 + 1, width]
+
+            standard_n_tokens = 50 if self.model.conv1.kernel_size[0] == 32 else 197
+
+            if x.shape[1] != standard_n_tokens:
+                new_shape = int(math.sqrt(x.shape[1]-1))
+                x = x + self.rescaled_pos_emb((new_shape, new_shape)).to(x.dtype)[None,:,:]
+            else:
+                x = x + self.model.positional_embedding.to(x.dtype)
+
+            x = self.model.ln_pre(x)
+
+            x = x.permute(1, 0, 2)  # NLD -> LND
+
+            activations, affinities = [], []
+            for i, res_block in enumerate(self.model.transformer.resblocks):
+                
+                if mask is not None:
+                    mask_layer, mask_type, mask_tensor = mask
+                    if mask_layer == i or mask_layer == 'all':
+                        # import ipdb; ipdb.set_trace()
+                        size = int(math.sqrt(x.shape[0] - 1))
+                        
+                        attn_mask = (mask_type, nnf.interpolate(mask_tensor.unsqueeze(1).float(), (size, size)).view(mask_tensor.shape[0], size * size))
+                        
+                    else:
+                        attn_mask = None
+                else:
+                    attn_mask = None
+
+                x, aff_per_head = forward_multihead_attention(x, res_block, with_aff=True, attn_mask=attn_mask)
+
+                if i in extract_layers:
+                    affinities += [aff_per_head]
+
+                    #if self.n_tokens is not None:
+                    #    activations += [nnf.interpolate(x, inp_size, mode='bilinear', align_corners=True)]
+                    #else:
+                    activations += [x]
+
+                if len(extract_layers) > 0 and i == max(extract_layers) and skip:
+                    print('early skip')
+                    break
+                
+            x = x.permute(1, 0, 2)  # LND -> NLD
+            x = self.model.ln_post(x[:, 0, :])
+
+            if self.model.proj is not None:
+                x = x @ self.model.proj
+
+            return x, activations, affinities
+
+    def sample_prompts(self, words, prompt_list=None):
+
+        prompt_list = prompt_list if prompt_list is not None else self.prompt_list
+
+        prompt_indices = torch.multinomial(torch.ones(len(prompt_list)), len(words), replacement=True)
+        prompts = [prompt_list[i] for i in prompt_indices]
+        return [promt.format(w) for promt, w in zip(prompts, words)]
+
+    def get_cond_vec(self, conditional, batch_size):
+        # compute conditional from a single string
+        if conditional is not None and type(conditional) == str:
+            cond = self.compute_conditional(conditional)
+            cond = cond.repeat(batch_size, 1)
+
+        # compute conditional from string list/tuple
+        elif conditional is not None and type(conditional) in {list, tuple} and type(conditional[0]) == str:
+            assert len(conditional) == batch_size
+            cond = self.compute_conditional(conditional)
+
+        # use conditional directly
+        elif conditional is not None and type(conditional) == torch.Tensor and conditional.ndim == 2:
+            cond = conditional
+
+        # compute conditional from image
+        elif conditional is not None and type(conditional) == torch.Tensor:
+            with torch.no_grad():
+                cond, _, _ = self.visual_forward(conditional)
+        else:
+            raise ValueError('invalid conditional')
+        return cond   
+
+    def compute_conditional(self, conditional):
+        import clip
+
+        dev = next(self.parameters()).device
+
+        if type(conditional) in {list, tuple}:
+            text_tokens = clip.tokenize(conditional).to(dev)
+            cond = self.clip_model.encode_text(text_tokens)
+        else:
+            if conditional in self.precomputed_prompts:
+                cond = self.precomputed_prompts[conditional].float().to(dev)
+            else:
+                text_tokens = clip.tokenize([conditional]).to(dev)
+                cond = self.clip_model.encode_text(text_tokens)[0]
+        
+        if self.shift_vector is not None:
+            return cond + self.shift_vector
+        else:
+            return cond
+
+
+def clip_load_untrained(version):
+    assert version == 'ViT-B/16'
+    from clip.model import CLIP
+    from clip.clip import _MODELS, _download
+    model = torch.jit.load(_download(_MODELS['ViT-B/16'])).eval()
+    state_dict = model.state_dict()
+
+    vision_width = state_dict["visual.conv1.weight"].shape[0]
+    vision_layers = len([k for k in state_dict.keys() if k.startswith("visual.") and k.endswith(".attn.in_proj_weight")])
+    vision_patch_size = state_dict["visual.conv1.weight"].shape[-1]
+    grid_size = round((state_dict["visual.positional_embedding"].shape[0] - 1) ** 0.5)
+    image_resolution = vision_patch_size * grid_size
+    embed_dim = state_dict["text_projection"].shape[1]
+    context_length = state_dict["positional_embedding"].shape[0]
+    vocab_size = state_dict["token_embedding.weight"].shape[0]
+    transformer_width = state_dict["ln_final.weight"].shape[0]
+    transformer_heads = transformer_width // 64
+    transformer_layers = len(set(k.split(".")[2] for k in state_dict if k.startswith(f"transformer.resblocks")))
+
+    return CLIP(embed_dim, image_resolution, vision_layers, vision_width, vision_patch_size, 
+        context_length, vocab_size, transformer_width, transformer_heads, transformer_layers)    
+
+
+class CLIPDensePredT(CLIPDenseBase):
+
+    def __init__(self, version='ViT-B/32', extract_layers=(3, 6, 9), cond_layer=0, reduce_dim=128, n_heads=4, prompt='fixed', 
+                 extra_blocks=0, reduce_cond=None, fix_shift=False,
+                 learn_trans_conv_only=False,  limit_to_clip_only=False, upsample=False, 
+                 add_calibration=False, rev_activations=False, trans_conv=None, n_tokens=None, complex_trans_conv=False):
+        
+        super().__init__(version, reduce_cond, reduce_dim, prompt, n_tokens)
+        # device = 'cpu'
+
+        self.extract_layers = extract_layers
+        self.cond_layer = cond_layer
+        self.limit_to_clip_only = limit_to_clip_only
+        self.process_cond = None
+        self.rev_activations = rev_activations
+        
+        depth = len(extract_layers)
+
+        if add_calibration:
+            self.calibration_conds = 1
+
+        self.upsample_proj = nn.Conv2d(reduce_dim, 1, kernel_size=1) if upsample else None
+
+        self.add_activation1 = True
+
+        self.version = version
+        
+        self.token_shape = {'ViT-B/32': (7, 7), 'ViT-B/16': (14, 14)}[version]
+
+        if fix_shift:
+            # self.shift_vector = nn.Parameter(torch.load(join(dirname(basename(__file__)), 'clip_text_shift_vector.pth')), requires_grad=False)
+            self.shift_vector = nn.Parameter(torch.load(join(dirname(basename(__file__)), 'shift_text_to_vis.pth')), requires_grad=False)
+            # self.shift_vector = nn.Parameter(-1*torch.load(join(dirname(basename(__file__)), 'shift2.pth')), requires_grad=False)
+        else:
+            self.shift_vector = None
+
+        if trans_conv is None:
+            trans_conv_ks = {'ViT-B/32': (32, 32), 'ViT-B/16': (16, 16)}[version]
+        else:
+            # explicitly define transposed conv kernel size
+            trans_conv_ks = (trans_conv, trans_conv)
+
+        if not complex_trans_conv:
+            self.trans_conv = nn.ConvTranspose2d(reduce_dim, 1, trans_conv_ks, stride=trans_conv_ks)
+        else:
+            assert trans_conv_ks[0] == trans_conv_ks[1]
+
+            tp_kernels = (trans_conv_ks[0] // 4, trans_conv_ks[0] // 4)
+
+            self.trans_conv = nn.Sequential(
+                nn.Conv2d(reduce_dim, reduce_dim, kernel_size=3, padding=1),
+                nn.ReLU(),
+                nn.ConvTranspose2d(reduce_dim, reduce_dim // 2, kernel_size=tp_kernels[0], stride=tp_kernels[0]),
+                nn.ReLU(),
+                nn.ConvTranspose2d(reduce_dim // 2, 1, kernel_size=tp_kernels[1], stride=tp_kernels[1]),               
+            )
+
+#        self.trans_conv = nn.ConvTranspose2d(reduce_dim, 1, trans_conv_ks, stride=trans_conv_ks)
+        
+        assert len(self.extract_layers) == depth
+
+        self.reduces = nn.ModuleList([nn.Linear(768, reduce_dim) for _ in range(depth)])
+        self.blocks = nn.ModuleList([nn.TransformerEncoderLayer(d_model=reduce_dim, nhead=n_heads) for _ in range(len(self.extract_layers))])
+        self.extra_blocks = nn.ModuleList([nn.TransformerEncoderLayer(d_model=reduce_dim, nhead=n_heads) for _ in range(extra_blocks)])
+        
+        # refinement and trans conv
+
+        if learn_trans_conv_only:
+            for p in self.parameters():
+                p.requires_grad_(False)
+            
+            for p in self.trans_conv.parameters():
+                p.requires_grad_(True)
+
+        self.prompt_list = get_prompt_list(prompt)
+
+
+    def forward(self, inp_image, conditional=None, return_features=False, mask=None):
+
+        assert type(return_features) == bool
+
+        inp_image = inp_image.to(self.model.positional_embedding.device)
+
+        if mask is not None:
+            raise ValueError('mask not supported')
+
+        # x_inp = normalize(inp_image)
+        x_inp = inp_image
+
+        bs, dev = inp_image.shape[0], x_inp.device
+
+        cond = self.get_cond_vec(conditional, bs)
+
+        visual_q, activations, _ = self.visual_forward(x_inp, extract_layers=[0] + list(self.extract_layers))
+
+        activation1 = activations[0]
+        activations = activations[1:]
+
+        _activations = activations[::-1] if not self.rev_activations else activations
+
+        a = None
+        for i, (activation, block, reduce) in enumerate(zip(_activations, self.blocks, self.reduces)):
+            
+            if a is not None:
+                a = reduce(activation) + a
+            else:
+                a = reduce(activation)
+
+            if i == self.cond_layer:
+                if self.reduce_cond is not None:
+                    cond = self.reduce_cond(cond)
+                
+                a = self.film_mul(cond) * a + self.film_add(cond)
+
+            a = block(a)
+
+        for block in self.extra_blocks:
+            a = a + block(a)
+
+        a = a[1:].permute(1, 2, 0) # rm cls token and -> BS, Feats, Tokens
+
+        size = int(math.sqrt(a.shape[2]))
+
+        a = a.view(bs, a.shape[1], size, size)
+
+        a = self.trans_conv(a)
+
+        if self.n_tokens is not None:
+            a = nnf.interpolate(a, x_inp.shape[2:], mode='bilinear', align_corners=True) 
+
+        if self.upsample_proj is not None:
+            a = self.upsample_proj(a)
+            a = nnf.interpolate(a, x_inp.shape[2:], mode='bilinear')
+
+        if return_features:
+            return a, visual_q, cond, [activation1] + activations
+        else:
+            return a,
+
+
+
+class CLIPDensePredTMasked(CLIPDensePredT):
+
+    def __init__(self, version='ViT-B/32', extract_layers=(3, 6, 9), cond_layer=0, reduce_dim=128, n_heads=4, 
+                 prompt='fixed', extra_blocks=0, reduce_cond=None, fix_shift=False, learn_trans_conv_only=False, 
+                 refine=None, limit_to_clip_only=False, upsample=False, add_calibration=False, n_tokens=None):
+
+        super().__init__(version=version, extract_layers=extract_layers, cond_layer=cond_layer, reduce_dim=reduce_dim, 
+                         n_heads=n_heads, prompt=prompt, extra_blocks=extra_blocks, reduce_cond=reduce_cond, 
+                         fix_shift=fix_shift, learn_trans_conv_only=learn_trans_conv_only,
+                         limit_to_clip_only=limit_to_clip_only, upsample=upsample, add_calibration=add_calibration,
+                         n_tokens=n_tokens)
+
+    def visual_forward_masked(self, img_s, seg_s):
+        return super().visual_forward(img_s, mask=('all', 'cls_token', seg_s))
+
+    def forward(self, img_q, cond_or_img_s, seg_s=None, return_features=False):
+
+        if seg_s is None:
+            cond = cond_or_img_s
+        else:
+            img_s = cond_or_img_s
+
+            with torch.no_grad():
+                cond, _, _ = self.visual_forward_masked(img_s, seg_s)
+
+        return super().forward(img_q, cond, return_features=return_features)
+
+
+
+class CLIPDenseBaseline(CLIPDenseBase):
+
+    def __init__(self, version='ViT-B/32', cond_layer=0, 
+                extract_layer=9, reduce_dim=128, reduce2_dim=None, prompt='fixed', 
+                 reduce_cond=None, limit_to_clip_only=False, n_tokens=None):
+        
+        super().__init__(version, reduce_cond, reduce_dim, prompt, n_tokens)
+        device = 'cpu'
+
+        # self.cond_layer = cond_layer
+        self.extract_layer = extract_layer
+        self.limit_to_clip_only = limit_to_clip_only
+        self.shift_vector = None
+
+        self.token_shape = {'ViT-B/32': (7, 7), 'ViT-B/16': (14, 14)}[version]
+        
+        assert reduce2_dim is not None
+
+        self.reduce2 = nn.Sequential(
+            nn.Linear(reduce_dim, reduce2_dim),
+            nn.ReLU(),
+            nn.Linear(reduce2_dim, reduce_dim)
+        )
+        
+        trans_conv_ks = {'ViT-B/32': (32, 32), 'ViT-B/16': (16, 16)}[version]
+        self.trans_conv = nn.ConvTranspose2d(reduce_dim, 1, trans_conv_ks, stride=trans_conv_ks)
+
+
+    def forward(self, inp_image, conditional=None, return_features=False):
+
+        inp_image = inp_image.to(self.model.positional_embedding.device)
+
+        # x_inp = normalize(inp_image)
+        x_inp = inp_image
+
+        bs, dev = inp_image.shape[0], x_inp.device
+
+        cond = self.get_cond_vec(conditional, bs)
+
+        visual_q, activations, affinities = self.visual_forward(x_inp, extract_layers=[self.extract_layer])
+
+        a = activations[0]
+        a = self.reduce(a)
+        a = self.film_mul(cond) * a + self.film_add(cond)
+
+        if self.reduce2 is not None:
+            a = self.reduce2(a)
+
+        # the original model would execute a transformer block here
+
+        a = a[1:].permute(1, 2, 0) # rm cls token and -> BS, Feats, Tokens
+
+        size = int(math.sqrt(a.shape[2]))
+
+        a = a.view(bs, a.shape[1], size, size)
+        a = self.trans_conv(a)
+
+        if return_features:
+            return a, visual_q, cond, activations
+        else:
+            return a,
+
+
+class CLIPSegMultiLabel(nn.Module):
+
+    def __init__(self, model) -> None:
+        super().__init__()
+
+        from third_party.JoEm.data_loader import get_seen_idx, get_unseen_idx, VOC
+
+        self.pascal_classes = VOC
+
+        from clip.clipseg import CLIPDensePredT
+        from general_utils import load_model
+        # self.clipseg = load_model('rd64-vit16-neg0.2-phrasecut', strict=False)
+        self.clipseg = load_model(model, strict=False)
+        
+        self.clipseg.eval()
+
+    def forward(self, x):
+
+        bs = x.shape[0]
+        out = torch.ones(21, bs, 352, 352).to(x.device) * -10
+
+        for class_id, class_name in enumerate(self.pascal_classes):
+        
+            fac = 3 if class_name == 'background' else 1
+
+            with torch.no_grad():
+                pred = torch.sigmoid(self.clipseg(x, class_name)[0][:,0]) * fac
+
+            out[class_id] += pred
+
+
+        out = out.permute(1, 0, 2, 3)
+
+        return out
+
+        # construct output tensor
+                    

clip/model.py

@@ -0,0 +1,436 @@
+from collections import OrderedDict
+from typing import Tuple, Union
+
+import numpy as np
+import torch
+import torch.nn.functional as F
+from torch import nn
+
+
+class Bottleneck(nn.Module):
+    expansion = 4
+
+    def __init__(self, inplanes, planes, stride=1):
+        super().__init__()
+
+        # all conv layers have stride 1. an avgpool is performed after the second convolution when stride > 1
+        self.conv1 = nn.Conv2d(inplanes, planes, 1, bias=False)
+        self.bn1 = nn.BatchNorm2d(planes)
+        self.relu1 = nn.ReLU(inplace=True)
+
+        self.conv2 = nn.Conv2d(planes, planes, 3, padding=1, bias=False)
+        self.bn2 = nn.BatchNorm2d(planes)
+        self.relu2 = nn.ReLU(inplace=True)
+
+        self.avgpool = nn.AvgPool2d(stride) if stride > 1 else nn.Identity()
+
+        self.conv3 = nn.Conv2d(planes, planes * self.expansion, 1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.relu3 = nn.ReLU(inplace=True)
+
+        self.downsample = None
+        self.stride = stride
+
+        if stride > 1 or inplanes != planes * Bottleneck.expansion:
+            # downsampling layer is prepended with an avgpool, and the subsequent convolution has stride 1
+            self.downsample = nn.Sequential(OrderedDict([
+                ("-1", nn.AvgPool2d(stride)),
+                ("0", nn.Conv2d(inplanes, planes * self.expansion, 1, stride=1, bias=False)),
+                ("1", nn.BatchNorm2d(planes * self.expansion))
+            ]))
+
+    def forward(self, x: torch.Tensor):
+        identity = x
+
+        out = self.relu1(self.bn1(self.conv1(x)))
+        out = self.relu2(self.bn2(self.conv2(out)))
+        out = self.avgpool(out)
+        out = self.bn3(self.conv3(out))
+
+        if self.downsample is not None:
+            identity = self.downsample(x)
+
+        out += identity
+        out = self.relu3(out)
+        return out
+
+
+class AttentionPool2d(nn.Module):
+    def __init__(self, spacial_dim: int, embed_dim: int, num_heads: int, output_dim: int = None):
+        super().__init__()
+        self.positional_embedding = nn.Parameter(torch.randn(spacial_dim ** 2 + 1, embed_dim) / embed_dim ** 0.5)
+        self.k_proj = nn.Linear(embed_dim, embed_dim)
+        self.q_proj = nn.Linear(embed_dim, embed_dim)
+        self.v_proj = nn.Linear(embed_dim, embed_dim)
+        self.c_proj = nn.Linear(embed_dim, output_dim or embed_dim)
+        self.num_heads = num_heads
+
+    def forward(self, x):
+        x = x.flatten(start_dim=2).permute(2, 0, 1)  # NCHW -> (HW)NC
+        x = torch.cat([x.mean(dim=0, keepdim=True), x], dim=0)  # (HW+1)NC
+        x = x + self.positional_embedding[:, None, :].to(x.dtype)  # (HW+1)NC
+        x, _ = F.multi_head_attention_forward(
+            query=x[:1], key=x, value=x,
+            embed_dim_to_check=x.shape[-1],
+            num_heads=self.num_heads,
+            q_proj_weight=self.q_proj.weight,
+            k_proj_weight=self.k_proj.weight,
+            v_proj_weight=self.v_proj.weight,
+            in_proj_weight=None,
+            in_proj_bias=torch.cat([self.q_proj.bias, self.k_proj.bias, self.v_proj.bias]),
+            bias_k=None,
+            bias_v=None,
+            add_zero_attn=False,
+            dropout_p=0,
+            out_proj_weight=self.c_proj.weight,
+            out_proj_bias=self.c_proj.bias,
+            use_separate_proj_weight=True,
+            training=self.training,
+            need_weights=False
+        )
+        return x.squeeze(0)
+
+
+class ModifiedResNet(nn.Module):
+    """
+    A ResNet class that is similar to torchvision's but contains the following changes:
+    - There are now 3 "stem" convolutions as opposed to 1, with an average pool instead of a max pool.
+    - Performs anti-aliasing strided convolutions, where an avgpool is prepended to convolutions with stride > 1
+    - The final pooling layer is a QKV attention instead of an average pool
+    """
+
+    def __init__(self, layers, output_dim, heads, input_resolution=224, width=64):
+        super().__init__()
+        self.output_dim = output_dim
+        self.input_resolution = input_resolution
+
+        # the 3-layer stem
+        self.conv1 = nn.Conv2d(3, width // 2, kernel_size=3, stride=2, padding=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(width // 2)
+        self.relu1 = nn.ReLU(inplace=True)
+        self.conv2 = nn.Conv2d(width // 2, width // 2, kernel_size=3, padding=1, bias=False)
+        self.bn2 = nn.BatchNorm2d(width // 2)
+        self.relu2 = nn.ReLU(inplace=True)
+        self.conv3 = nn.Conv2d(width // 2, width, kernel_size=3, padding=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(width)
+        self.relu3 = nn.ReLU(inplace=True)
+        self.avgpool = nn.AvgPool2d(2)
+
+        # residual layers
+        self._inplanes = width  # this is a *mutable* variable used during construction
+        self.layer1 = self._make_layer(width, layers[0])
+        self.layer2 = self._make_layer(width * 2, layers[1], stride=2)
+        self.layer3 = self._make_layer(width * 4, layers[2], stride=2)
+        self.layer4 = self._make_layer(width * 8, layers[3], stride=2)
+
+        embed_dim = width * 32  # the ResNet feature dimension
+        self.attnpool = AttentionPool2d(input_resolution // 32, embed_dim, heads, output_dim)
+
+    def _make_layer(self, planes, blocks, stride=1):
+        layers = [Bottleneck(self._inplanes, planes, stride)]
+
+        self._inplanes = planes * Bottleneck.expansion
+        for _ in range(1, blocks):
+            layers.append(Bottleneck(self._inplanes, planes))
+
+        return nn.Sequential(*layers)
+
+    def forward(self, x):
+        def stem(x):
+            x = self.relu1(self.bn1(self.conv1(x)))
+            x = self.relu2(self.bn2(self.conv2(x)))
+            x = self.relu3(self.bn3(self.conv3(x)))
+            x = self.avgpool(x)
+            return x
+
+        x = x.type(self.conv1.weight.dtype)
+        x = stem(x)
+        x = self.layer1(x)
+        x = self.layer2(x)
+        x = self.layer3(x)
+        x = self.layer4(x)
+        x = self.attnpool(x)
+
+        return x
+
+
+class LayerNorm(nn.LayerNorm):
+    """Subclass torch's LayerNorm to handle fp16."""
+
+    def forward(self, x: torch.Tensor):
+        orig_type = x.dtype
+        ret = super().forward(x.type(torch.float32))
+        return ret.type(orig_type)
+
+
+class QuickGELU(nn.Module):
+    def forward(self, x: torch.Tensor):
+        return x * torch.sigmoid(1.702 * x)
+
+
+class ResidualAttentionBlock(nn.Module):
+    def __init__(self, d_model: int, n_head: int, attn_mask: torch.Tensor = None):
+        super().__init__()
+
+        self.attn = nn.MultiheadAttention(d_model, n_head)
+        self.ln_1 = LayerNorm(d_model)
+        self.mlp = nn.Sequential(OrderedDict([
+            ("c_fc", nn.Linear(d_model, d_model * 4)),
+            ("gelu", QuickGELU()),
+            ("c_proj", nn.Linear(d_model * 4, d_model))
+        ]))
+        self.ln_2 = LayerNorm(d_model)
+        self.attn_mask = attn_mask
+
+    def attention(self, x: torch.Tensor):
+        self.attn_mask = self.attn_mask.to(dtype=x.dtype, device=x.device) if self.attn_mask is not None else None
+        return self.attn(x, x, x, need_weights=False, attn_mask=self.attn_mask)[0]
+
+    def forward(self, x: torch.Tensor):
+        x = x + self.attention(self.ln_1(x))
+        x = x + self.mlp(self.ln_2(x))
+        return x
+
+
+class Transformer(nn.Module):
+    def __init__(self, width: int, layers: int, heads: int, attn_mask: torch.Tensor = None):
+        super().__init__()
+        self.width = width
+        self.layers = layers
+        self.resblocks = nn.Sequential(*[ResidualAttentionBlock(width, heads, attn_mask) for _ in range(layers)])
+
+    def forward(self, x: torch.Tensor):
+        return self.resblocks(x)
+
+
+class VisionTransformer(nn.Module):
+    def __init__(self, input_resolution: int, patch_size: int, width: int, layers: int, heads: int, output_dim: int):
+        super().__init__()
+        self.input_resolution = input_resolution
+        self.output_dim = output_dim
+        self.conv1 = nn.Conv2d(in_channels=3, out_channels=width, kernel_size=patch_size, stride=patch_size, bias=False)
+
+        scale = width ** -0.5
+        self.class_embedding = nn.Parameter(scale * torch.randn(width))
+        self.positional_embedding = nn.Parameter(scale * torch.randn((input_resolution // patch_size) ** 2 + 1, width))
+        self.ln_pre = LayerNorm(width)
+
+        self.transformer = Transformer(width, layers, heads)
+
+        self.ln_post = LayerNorm(width)
+        self.proj = nn.Parameter(scale * torch.randn(width, output_dim))
+
+    def forward(self, x: torch.Tensor):
+        x = self.conv1(x)  # shape = [*, width, grid, grid]
+        x = x.reshape(x.shape[0], x.shape[1], -1)  # shape = [*, width, grid ** 2]
+        x = x.permute(0, 2, 1)  # shape = [*, grid ** 2, width]
+        x = torch.cat([self.class_embedding.to(x.dtype) + torch.zeros(x.shape[0], 1, x.shape[-1], dtype=x.dtype, device=x.device), x], dim=1)  # shape = [*, grid ** 2 + 1, width]
+        x = x + self.positional_embedding.to(x.dtype)
+        x = self.ln_pre(x)
+
+        x = x.permute(1, 0, 2)  # NLD -> LND
+        x = self.transformer(x)
+        x = x.permute(1, 0, 2)  # LND -> NLD
+
+        x = self.ln_post(x[:, 0, :])
+
+        if self.proj is not None:
+            x = x @ self.proj
+
+        return x
+
+
+class CLIP(nn.Module):
+    def __init__(self,
+                 embed_dim: int,
+                 # vision
+                 image_resolution: int,
+                 vision_layers: Union[Tuple[int, int, int, int], int],
+                 vision_width: int,
+                 vision_patch_size: int,
+                 # text
+                 context_length: int,
+                 vocab_size: int,
+                 transformer_width: int,
+                 transformer_heads: int,
+                 transformer_layers: int
+                 ):
+        super().__init__()
+
+        self.context_length = context_length
+
+        if isinstance(vision_layers, (tuple, list)):
+            vision_heads = vision_width * 32 // 64
+            self.visual = ModifiedResNet(
+                layers=vision_layers,
+                output_dim=embed_dim,
+                heads=vision_heads,
+                input_resolution=image_resolution,
+                width=vision_width
+            )
+        else:
+            vision_heads = vision_width // 64
+            self.visual = VisionTransformer(
+                input_resolution=image_resolution,
+                patch_size=vision_patch_size,
+                width=vision_width,
+                layers=vision_layers,
+                heads=vision_heads,
+                output_dim=embed_dim
+            )
+
+        self.transformer = Transformer(
+            width=transformer_width,
+            layers=transformer_layers,
+            heads=transformer_heads,
+            attn_mask=self.build_attention_mask()
+        )
+
+        self.vocab_size = vocab_size
+        self.token_embedding = nn.Embedding(vocab_size, transformer_width)
+        self.positional_embedding = nn.Parameter(torch.empty(self.context_length, transformer_width))
+        self.ln_final = LayerNorm(transformer_width)
+
+        self.text_projection = nn.Parameter(torch.empty(transformer_width, embed_dim))
+        self.logit_scale = nn.Parameter(torch.ones([]) * np.log(1 / 0.07))
+
+        self.initialize_parameters()
+
+    def initialize_parameters(self):
+        nn.init.normal_(self.token_embedding.weight, std=0.02)
+        nn.init.normal_(self.positional_embedding, std=0.01)
+
+        if isinstance(self.visual, ModifiedResNet):
+            if self.visual.attnpool is not None:
+                std = self.visual.attnpool.c_proj.in_features ** -0.5
+                nn.init.normal_(self.visual.attnpool.q_proj.weight, std=std)
+                nn.init.normal_(self.visual.attnpool.k_proj.weight, std=std)
+                nn.init.normal_(self.visual.attnpool.v_proj.weight, std=std)
+                nn.init.normal_(self.visual.attnpool.c_proj.weight, std=std)
+
+            for resnet_block in [self.visual.layer1, self.visual.layer2, self.visual.layer3, self.visual.layer4]:
+                for name, param in resnet_block.named_parameters():
+                    if name.endswith("bn3.weight"):
+                        nn.init.zeros_(param)
+
+        proj_std = (self.transformer.width ** -0.5) * ((2 * self.transformer.layers) ** -0.5)
+        attn_std = self.transformer.width ** -0.5
+        fc_std = (2 * self.transformer.width) ** -0.5
+        for block in self.transformer.resblocks:
+            nn.init.normal_(block.attn.in_proj_weight, std=attn_std)
+            nn.init.normal_(block.attn.out_proj.weight, std=proj_std)
+            nn.init.normal_(block.mlp.c_fc.weight, std=fc_std)
+            nn.init.normal_(block.mlp.c_proj.weight, std=proj_std)
+
+        if self.text_projection is not None:
+            nn.init.normal_(self.text_projection, std=self.transformer.width ** -0.5)
+
+    def build_attention_mask(self):
+        # lazily create causal attention mask, with full attention between the vision tokens
+        # pytorch uses additive attention mask; fill with -inf
+        mask = torch.empty(self.context_length, self.context_length)
+        mask.fill_(float("-inf"))
+        mask.triu_(1)  # zero out the lower diagonal
+        return mask
+
+    @property
+    def dtype(self):
+        return self.visual.conv1.weight.dtype
+
+    def encode_image(self, image):
+        return self.visual(image.type(self.dtype))
+
+    def encode_text(self, text):
+        x = self.token_embedding(text).type(self.dtype)  # [batch_size, n_ctx, d_model]
+
+        x = x + self.positional_embedding.type(self.dtype)
+        x = x.permute(1, 0, 2)  # NLD -> LND
+        x = self.transformer(x)
+        x = x.permute(1, 0, 2)  # LND -> NLD
+        x = self.ln_final(x).type(self.dtype)
+
+        # x.shape = [batch_size, n_ctx, transformer.width]
+        # take features from the eot embedding (eot_token is the highest number in each sequence)
+        x = x[torch.arange(x.shape[0]), text.argmax(dim=-1)] @ self.text_projection
+
+        return x
+
+    def forward(self, image, text):
+        image_features = self.encode_image(image)
+        text_features = self.encode_text(text)
+
+        # normalized features
+        image_features = image_features / image_features.norm(dim=1, keepdim=True)
+        text_features = text_features / text_features.norm(dim=1, keepdim=True)
+
+        # cosine similarity as logits
+        logit_scale = self.logit_scale.exp()
+        logits_per_image = logit_scale * image_features @ text_features.t()
+        logits_per_text = logits_per_image.t()
+
+        # shape = [global_batch_size, global_batch_size]
+        return logits_per_image, logits_per_text
+
+
+def convert_weights(model: nn.Module):
+    """Convert applicable model parameters to fp16"""
+
+    def _convert_weights_to_fp16(l):
+        if isinstance(l, (nn.Conv1d, nn.Conv2d, nn.Linear)):
+            l.weight.data = l.weight.data.half()
+            if l.bias is not None:
+                l.bias.data = l.bias.data.half()
+
+        if isinstance(l, nn.MultiheadAttention):
+            for attr in [*[f"{s}_proj_weight" for s in ["in", "q", "k", "v"]], "in_proj_bias", "bias_k", "bias_v"]:
+                tensor = getattr(l, attr)
+                if tensor is not None:
+                    tensor.data = tensor.data.half()
+
+        for name in ["text_projection", "proj"]:
+            if hasattr(l, name):
+                attr = getattr(l, name)
+                if attr is not None:
+                    attr.data = attr.data.half()
+
+    model.apply(_convert_weights_to_fp16)
+
+
+def build_model(state_dict: dict):
+    vit = "visual.proj" in state_dict
+
+    if vit:
+        vision_width = state_dict["visual.conv1.weight"].shape[0]
+        vision_layers = len([k for k in state_dict.keys() if k.startswith("visual.") and k.endswith(".attn.in_proj_weight")])
+        vision_patch_size = state_dict["visual.conv1.weight"].shape[-1]
+        grid_size = round((state_dict["visual.positional_embedding"].shape[0] - 1) ** 0.5)
+        image_resolution = vision_patch_size * grid_size
+    else:
+        counts: list = [len(set(k.split(".")[2] for k in state_dict if k.startswith(f"visual.layer{b}"))) for b in [1, 2, 3, 4]]
+        vision_layers = tuple(counts)
+        vision_width = state_dict["visual.layer1.0.conv1.weight"].shape[0]
+        output_width = round((state_dict["visual.attnpool.positional_embedding"].shape[0] - 1) ** 0.5)
+        vision_patch_size = None
+        assert output_width ** 2 + 1 == state_dict["visual.attnpool.positional_embedding"].shape[0]
+        image_resolution = output_width * 32
+
+    embed_dim = state_dict["text_projection"].shape[1]
+    context_length = state_dict["positional_embedding"].shape[0]
+    vocab_size = state_dict["token_embedding.weight"].shape[0]
+    transformer_width = state_dict["ln_final.weight"].shape[0]
+    transformer_heads = transformer_width // 64
+    transformer_layers = len(set(k.split(".")[2] for k in state_dict if k.startswith("transformer.resblocks")))
+
+    model = CLIP(
+        embed_dim,
+        image_resolution, vision_layers, vision_width, vision_patch_size,
+        context_length, vocab_size, transformer_width, transformer_heads, transformer_layers
+    )
+
+    for key in ["input_resolution", "context_length", "vocab_size"]:
+        if key in state_dict:
+            del state_dict[key]
+
+    convert_weights(model)
+    model.load_state_dict(state_dict)
+    return model.eval()

clip/simple_tokenizer.py

@@ -0,0 +1,132 @@
+import gzip
+import html
+import os
+from functools import lru_cache
+
+import ftfy
+import regex as re
+
+
+@lru_cache()
+def default_bpe():
+    return os.path.join(os.path.dirname(os.path.abspath(__file__)), "bpe_simple_vocab_16e6.txt.gz")
+
+
+@lru_cache()
+def bytes_to_unicode():
+    """
+    Returns list of utf-8 byte and a corresponding list of unicode strings.
+    The reversible bpe codes work on unicode strings.
+    This means you need a large # of unicode characters in your vocab if you want to avoid UNKs.
+    When you're at something like a 10B token dataset you end up needing around 5K for decent coverage.
+    This is a signficant percentage of your normal, say, 32K bpe vocab.
+    To avoid that, we want lookup tables between utf-8 bytes and unicode strings.
+    And avoids mapping to whitespace/control characters the bpe code barfs on.
+    """
+    bs = list(range(ord("!"), ord("~")+1))+list(range(ord("¡"), ord("¬")+1))+list(range(ord("®"), ord("ÿ")+1))
+    cs = bs[:]
+    n = 0
+    for b in range(2**8):
+        if b not in bs:
+            bs.append(b)
+            cs.append(2**8+n)
+            n += 1
+    cs = [chr(n) for n in cs]
+    return dict(zip(bs, cs))
+
+
+def get_pairs(word):
+    """Return set of symbol pairs in a word.
+    Word is represented as tuple of symbols (symbols being variable-length strings).
+    """
+    pairs = set()
+    prev_char = word[0]
+    for char in word[1:]:
+        pairs.add((prev_char, char))
+        prev_char = char
+    return pairs
+
+
+def basic_clean(text):
+    text = ftfy.fix_text(text)
+    text = html.unescape(html.unescape(text))
+    return text.strip()
+
+
+def whitespace_clean(text):
+    text = re.sub(r'\s+', ' ', text)
+    text = text.strip()
+    return text
+
+
+class SimpleTokenizer(object):
+    def __init__(self, bpe_path: str = default_bpe()):
+        self.byte_encoder = bytes_to_unicode()
+        self.byte_decoder = {v: k for k, v in self.byte_encoder.items()}
+        merges = gzip.open(bpe_path).read().decode("utf-8").split('\n')
+        merges = merges[1:49152-256-2+1]
+        merges = [tuple(merge.split()) for merge in merges]
+        vocab = list(bytes_to_unicode().values())
+        vocab = vocab + [v+'</w>' for v in vocab]
+        for merge in merges:
+            vocab.append(''.join(merge))
+        vocab.extend(['<|startoftext|>', '<|endoftext|>'])
+        self.encoder = dict(zip(vocab, range(len(vocab))))
+        self.decoder = {v: k for k, v in self.encoder.items()}
+        self.bpe_ranks = dict(zip(merges, range(len(merges))))
+        self.cache = {'<|startoftext|>': '<|startoftext|>', '<|endoftext|>': '<|endoftext|>'}
+        self.pat = re.compile(r"""<\|startoftext\|>|<\|endoftext\|>|'s|'t|'re|'ve|'m|'ll|'d|[\p{L}]+|[\p{N}]|[^\s\p{L}\p{N}]+""", re.IGNORECASE)
+
+    def bpe(self, token):
+        if token in self.cache:
+            return self.cache[token]
+        word = tuple(token[:-1]) + ( token[-1] + '</w>',)
+        pairs = get_pairs(word)
+
+        if not pairs:
+            return token+'</w>'
+
+        while True:
+            bigram = min(pairs, key = lambda pair: self.bpe_ranks.get(pair, float('inf')))
+            if bigram not in self.bpe_ranks:
+                break
+            first, second = bigram
+            new_word = []
+            i = 0
+            while i < len(word):
+                try:
+                    j = word.index(first, i)
+                    new_word.extend(word[i:j])
+                    i = j
+                except:
+                    new_word.extend(word[i:])
+                    break
+
+                if word[i] == first and i < len(word)-1 and word[i+1] == second:
+                    new_word.append(first+second)
+                    i += 2
+                else:
+                    new_word.append(word[i])
+                    i += 1
+            new_word = tuple(new_word)
+            word = new_word
+            if len(word) == 1:
+                break
+            else:
+                pairs = get_pairs(word)
+        word = ' '.join(word)
+        self.cache[token] = word
+        return word
+
+    def encode(self, text):
+        bpe_tokens = []
+        text = whitespace_clean(basic_clean(text)).lower()
+        for token in re.findall(self.pat, text):
+            token = ''.join(self.byte_encoder[b] for b in token.encode('utf-8'))
+            bpe_tokens.extend(self.encoder[bpe_token] for bpe_token in self.bpe(token).split(' '))
+        return bpe_tokens
+
+    def decode(self, tokens):
+        text = ''.join([self.decoder[token] for token in tokens])
+        text = bytearray([self.byte_decoder[c] for c in text]).decode('utf-8', errors="replace").replace('</w>', ' ')
+        return text

clip/vitseg.py

@@ -0,0 +1,286 @@
+import math
+from posixpath import basename, dirname, join
+# import clip
+from clip.model import convert_weights
+import torch
+import json
+from torch import nn
+from torch.nn import functional as nnf
+from torch.nn.modules import activation
+from torch.nn.modules.activation import ReLU
+from torchvision import transforms
+
+normalize = transforms.Normalize(mean=(0.48145466, 0.4578275, 0.40821073), std=(0.26862954, 0.26130258, 0.27577711))
+
+from torchvision.models import ResNet
+
+
+def process_prompts(conditional, prompt_list, conditional_map):
+    # DEPRECATED
+            
+    # randomly sample a synonym
+    words = [conditional_map[int(i)] for i in conditional]
+    words = [syns[torch.multinomial(torch.ones(len(syns)), 1, replacement=True).item()] for syns in words]
+    words = [w.replace('_', ' ') for w in words]
+
+    if prompt_list is not None:
+        prompt_indices = torch.multinomial(torch.ones(len(prompt_list)), len(words), replacement=True)
+        prompts = [prompt_list[i] for i in prompt_indices]
+    else:
+        prompts = ['a photo of {}'] * (len(words))
+
+    return [promt.format(w) for promt, w in zip(prompts, words)]
+
+
+class VITDenseBase(nn.Module):
+    
+    def rescaled_pos_emb(self, new_size):
+        assert len(new_size) == 2
+
+        a = self.model.positional_embedding[1:].T.view(1, 768, *self.token_shape)
+        b = nnf.interpolate(a, new_size, mode='bicubic', align_corners=False).squeeze(0).view(768, new_size[0]*new_size[1]).T
+        return torch.cat([self.model.positional_embedding[:1], b])
+
+    def visual_forward(self, x_inp, extract_layers=(), skip=False, mask=None):
+        
+        with torch.no_grad():
+
+            x_inp = nnf.interpolate(x_inp, (384, 384))
+
+            x = self.model.patch_embed(x_inp)
+            cls_token = self.model.cls_token.expand(x.shape[0], -1, -1)  # stole cls_tokens impl from Phil Wang, thanks
+            if self.model.dist_token is None:
+                x = torch.cat((cls_token, x), dim=1)
+            else:
+                x = torch.cat((cls_token, self.model.dist_token.expand(x.shape[0], -1, -1), x), dim=1)
+            x = self.model.pos_drop(x + self.model.pos_embed)
+
+            activations = []
+            for i, block in enumerate(self.model.blocks):
+                x = block(x)
+
+                if i in extract_layers:
+                    # permute to be compatible with CLIP
+                    activations += [x.permute(1,0,2)]                
+
+            x = self.model.norm(x)
+            x = self.model.head(self.model.pre_logits(x[:, 0]))
+
+            # again for CLIP compatibility
+            # x = x.permute(1, 0, 2)
+
+        return x, activations, None
+
+    def sample_prompts(self, words, prompt_list=None):
+
+        prompt_list = prompt_list if prompt_list is not None else self.prompt_list
+
+        prompt_indices = torch.multinomial(torch.ones(len(prompt_list)), len(words), replacement=True)
+        prompts = [prompt_list[i] for i in prompt_indices]
+        return [promt.format(w) for promt, w in zip(prompts, words)]
+
+    def get_cond_vec(self, conditional, batch_size):
+        # compute conditional from a single string
+        if conditional is not None and type(conditional) == str:
+            cond = self.compute_conditional(conditional)
+            cond = cond.repeat(batch_size, 1)
+
+        # compute conditional from string list/tuple
+        elif conditional is not None and type(conditional) in {list, tuple} and type(conditional[0]) == str:
+            assert len(conditional) == batch_size
+            cond = self.compute_conditional(conditional)
+
+        # use conditional directly
+        elif conditional is not None and type(conditional) == torch.Tensor and conditional.ndim == 2:
+            cond = conditional
+
+        # compute conditional from image
+        elif conditional is not None and type(conditional) == torch.Tensor:
+            with torch.no_grad():
+                cond, _, _ = self.visual_forward(conditional)
+        else:
+            raise ValueError('invalid conditional')
+        return cond   
+
+    def compute_conditional(self, conditional):
+        import clip
+
+        dev = next(self.parameters()).device
+
+        if type(conditional) in {list, tuple}:
+            text_tokens = clip.tokenize(conditional).to(dev)
+            cond = self.clip_model.encode_text(text_tokens)
+        else:
+            if conditional in self.precomputed_prompts:
+                cond = self.precomputed_prompts[conditional].float().to(dev)
+            else:
+                text_tokens = clip.tokenize([conditional]).to(dev)
+                cond = self.clip_model.encode_text(text_tokens)[0]
+        
+        return cond
+
+
+class VITDensePredT(VITDenseBase):
+
+    def __init__(self, extract_layers=(3, 6, 9), cond_layer=0, reduce_dim=128, n_heads=4, prompt='fixed', 
+                 depth=3, extra_blocks=0, reduce_cond=None, fix_shift=False,
+                 learn_trans_conv_only=False, refine=None, limit_to_clip_only=False, upsample=False, 
+                 add_calibration=False, process_cond=None, not_pretrained=False):
+        super().__init__()
+        # device = 'cpu'
+
+        self.extract_layers = extract_layers
+        self.cond_layer = cond_layer
+        self.limit_to_clip_only = limit_to_clip_only
+        self.process_cond = None
+        
+        if add_calibration:
+            self.calibration_conds = 1
+
+        self.upsample_proj = nn.Conv2d(reduce_dim, 1, kernel_size=1) if upsample else None
+
+        self.add_activation1 = True
+
+        import timm 
+        self.model = timm.create_model('vit_base_patch16_384', pretrained=True)
+        self.model.head = nn.Linear(768, 512 if reduce_cond is None else reduce_cond)
+
+        for p in self.model.parameters():
+            p.requires_grad_(False)
+
+        import clip
+        self.clip_model, _ = clip.load('ViT-B/16', device='cpu', jit=False)
+        # del self.clip_model.visual
+        
+        
+        self.token_shape = (14, 14)
+
+        # conditional
+        if reduce_cond is not None:
+            self.reduce_cond = nn.Linear(512, reduce_cond)
+            for p in self.reduce_cond.parameters():
+                p.requires_grad_(False)
+        else:
+            self.reduce_cond = None
+
+        # self.film = AVAILABLE_BLOCKS['film'](512, 128)
+        self.film_mul = nn.Linear(512 if reduce_cond is None else reduce_cond, reduce_dim)
+        self.film_add = nn.Linear(512 if reduce_cond is None else reduce_cond, reduce_dim)
+        
+        # DEPRECATED
+        # self.conditional_map = {c['id']: c['synonyms'] for c in json.load(open(cond_map))}
+        
+        assert len(self.extract_layers) == depth
+
+        self.reduces = nn.ModuleList([nn.Linear(768, reduce_dim) for _ in range(depth)])
+        self.blocks = nn.ModuleList([nn.TransformerEncoderLayer(d_model=reduce_dim, nhead=n_heads) for _ in range(len(self.extract_layers))])
+        self.extra_blocks = nn.ModuleList([nn.TransformerEncoderLayer(d_model=reduce_dim, nhead=n_heads) for _ in range(extra_blocks)])
+
+        trans_conv_ks = (16, 16)
+        self.trans_conv = nn.ConvTranspose2d(reduce_dim, 1, trans_conv_ks, stride=trans_conv_ks)
+
+        # refinement and trans conv
+
+        if learn_trans_conv_only:
+            for p in self.parameters():
+                p.requires_grad_(False)
+            
+            for p in self.trans_conv.parameters():
+                p.requires_grad_(True)
+
+        if prompt == 'fixed':
+            self.prompt_list = ['a photo of a {}.']
+        elif prompt == 'shuffle':
+            self.prompt_list = ['a photo of a {}.', 'a photograph of a {}.', 'an image of a {}.', '{}.']
+        elif prompt == 'shuffle+':
+            self.prompt_list = ['a photo of a {}.', 'a photograph of a {}.', 'an image of a {}.', '{}.',
+                                'a cropped photo of a {}.', 'a good photo of a {}.', 'a photo of one {}.',
+                                'a bad photo of a {}.', 'a photo of the {}.']
+        elif prompt == 'shuffle_clip':
+            from models.clip_prompts import imagenet_templates
+            self.prompt_list = imagenet_templates
+
+        if process_cond is not None:
+            if process_cond == 'clamp' or process_cond[0] == 'clamp':
+
+                val = process_cond[1] if type(process_cond) in {list, tuple} else 0.2
+
+                def clamp_vec(x):
+                    return torch.clamp(x, -val, val)
+
+                self.process_cond = clamp_vec
+
+            elif process_cond.endswith('.pth'):
+                
+                shift = torch.load(process_cond)
+                def add_shift(x):
+                    return x + shift.to(x.device)
+
+                self.process_cond = add_shift
+
+        import pickle
+        precomp = pickle.load(open('precomputed_prompt_vectors.pickle', 'rb'))
+        self.precomputed_prompts = {k: torch.from_numpy(v) for k, v in precomp.items()}
+
+
+    def forward(self, inp_image, conditional=None, return_features=False, mask=None):
+
+        assert type(return_features) == bool
+
+        # inp_image = inp_image.to(self.model.positional_embedding.device)
+
+        if mask is not None:
+            raise ValueError('mask not supported')
+
+        # x_inp = normalize(inp_image)
+        x_inp = inp_image
+
+        bs, dev = inp_image.shape[0], x_inp.device
+
+        inp_image_size = inp_image.shape[2:]
+
+        cond = self.get_cond_vec(conditional, bs)
+
+        visual_q, activations, _ = self.visual_forward(x_inp, extract_layers=[0] + list(self.extract_layers))
+
+        activation1 = activations[0]
+        activations = activations[1:]
+
+        a = None
+        for i, (activation, block, reduce) in enumerate(zip(activations[::-1], self.blocks, self.reduces)):
+            
+            if a is not None:
+                a = reduce(activation) + a
+            else:
+                a = reduce(activation)
+
+            if i == self.cond_layer:
+                if self.reduce_cond is not None:
+                    cond = self.reduce_cond(cond)
+                
+                a = self.film_mul(cond) * a + self.film_add(cond)
+
+            a = block(a)
+
+        for block in self.extra_blocks:
+            a = a + block(a)
+
+        a = a[1:].permute(1, 2, 0) # rm cls token and -> BS, Feats, Tokens
+
+        size = int(math.sqrt(a.shape[2]))
+
+        a = a.view(bs, a.shape[1], size, size)
+
+        if self.trans_conv is not None:
+            a = self.trans_conv(a)
+
+        if self.upsample_proj is not None:
+            a = self.upsample_proj(a)
+            a = nnf.interpolate(a, x_inp.shape[2:], mode='bilinear')
+
+        a = nnf.interpolate(a, inp_image_size)
+
+        if return_features:
+            return a, visual_q, cond, [activation1] + activations
+        else:
+            return a,

gfpgan/weights/detection_Resnet50_Final.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d1de9c2944f2ccddca5f5e010ea5ae64a39845a86311af6fdf30841b0a5a16d
+size 109497761

gfpgan/weights/parsing_parsenet.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d558d8d0e42c20224f13cf5a29c79eba2d59913419f945545d8cf7b72920de2
+size 85331193

installer/installer.py

@@ -0,0 +1,83 @@
+import argparse
+import glob
+import os
+import shutil
+import site
+import subprocess
+import sys
+
+
+script_dir = os.getcwd()
+
+
+def run_cmd(cmd, capture_output=False, env=None):
+    # Run shell commands
+    return subprocess.run(cmd, shell=True, capture_output=capture_output, env=env)
+
+
+def check_env():
+    # If we have access to conda, we are probably in an environment
+    conda_not_exist = run_cmd("conda", capture_output=True).returncode
+    if conda_not_exist:
+        print("Conda is not installed. Exiting...")
+        sys.exit()
+    
+    # Ensure this is a new environment and not the base environment
+    if os.environ["CONDA_DEFAULT_ENV"] == "base":
+        print("Create an environment for this project and activate it. Exiting...")
+        sys.exit()
+
+
+def install_dependencies():
+    # Install Git and clone repo
+    run_cmd("conda install -y -k git")
+    run_cmd("git clone https://github.com/C0untFloyd/roop-unleashed.git")
+
+    # Install the webui dependencies
+    update_dependencies()
+
+
+def update_dependencies():
+    global MY_PATH
+    
+    os.chdir(MY_PATH)
+	# do a hard reset for to update even if there are local changes
+    run_cmd("git fetch --all")
+    run_cmd("git reset --hard origin/main")
+    run_cmd("git pull")
+    # Installs/Updates dependencies from all requirements.txt
+    run_cmd("python -m pip install -r requirements.txt")
+
+
+def start_app():
+    global MY_PATH
+    
+    os.chdir(MY_PATH)
+    # forward commandline arguments
+    sys.argv.pop(0)
+    args = ' '.join(sys.argv)
+    print("Launching App")
+    run_cmd(f'python run.py {args}')
+
+
+if __name__ == "__main__":
+    global MY_PATH
+    
+    MY_PATH = "roop-unleashed"
+
+    
+    # Verifies we are in a conda environment
+    check_env()
+
+    # If webui has already been installed, skip and run
+    if not os.path.exists(MY_PATH):
+        install_dependencies()
+    else:
+        # moved update from batch to here, because of batch limitations
+        updatechoice = input("Check for Updates? [y/n]").lower()
+        if updatechoice == "y":
+           update_dependencies()
+
+    # Run the model with webui
+    os.chdir(script_dir)
+    start_app()

installer/windows_run.bat

@@ -0,0 +1,80 @@
+@echo off
+REM Please set the following commandline arguments to your prefered settings
+set COMMANDLINE_ARGS=--execution-provider cuda --frame-processor face_swapper face_enhancer --video-encoder libvpx-vp9
+
+cd /D "%~dp0"
+
+echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniconda which can not be silently installed under a path with spaces. && goto end
+
+set PATH=%PATH%;%SystemRoot%\system32
+
+@rem config
+set INSTALL_DIR=%cd%\installer_files
+set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
+set INSTALL_ENV_DIR=%cd%\installer_files\env
+set MINICONDA_DOWNLOAD_URL=https://repo.anaconda.com/miniconda/Miniconda3-latest-Windows-x86_64.exe
+set FFMPEG_DOWNLOAD_URL=https://github.com/GyanD/codexffmpeg/releases/download/2023-06-21-git-1bcb8a7338/ffmpeg-2023-06-21-git-1bcb8a7338-essentials_build.zip
+set INSTALL_FFMPEG_DIR=%cd%\installer_files\ffmpeg
+set conda_exists=F
+
+@rem figure out whether git and conda needs to be installed
+call "%CONDA_ROOT_PREFIX%\_conda.exe" --version >nul 2>&1
+if "%ERRORLEVEL%" EQU "0" set conda_exists=T
+
+@rem (if necessary) install git and conda into a contained environment
+@rem download conda
+if "%conda_exists%" == "F" (
+	echo Downloading Miniconda from %MINICONDA_DOWNLOAD_URL% to %INSTALL_DIR%\miniconda_installer.exe
+
+	mkdir "%INSTALL_DIR%"
+	call curl -Lk "%MINICONDA_DOWNLOAD_URL%" > "%INSTALL_DIR%\miniconda_installer.exe" || ( echo. && echo Miniconda failed to download. && goto end )
+
+	echo Installing Miniconda to %CONDA_ROOT_PREFIX%
+	start /wait "" "%INSTALL_DIR%\miniconda_installer.exe" /InstallationType=JustMe /NoShortcuts=1 /AddToPath=0 /RegisterPython=0 /NoRegistry=1 /S /D=%CONDA_ROOT_PREFIX%
+
+	@rem test the conda binary
+	echo Miniconda version:
+	call "%CONDA_ROOT_PREFIX%\_conda.exe" --version || ( echo. && echo Miniconda not found. && goto end )
+)
+
+@rem create the installer env
+if not exist "%INSTALL_ENV_DIR%" (
+  echo Packages to install: %PACKAGES_TO_INSTALL%
+  call "%CONDA_ROOT_PREFIX%\_conda.exe" create --no-shortcuts -y -k --prefix "%INSTALL_ENV_DIR%" python=3.10 || ( echo. && echo Conda environment creation failed. && goto end )
+)
+
+if not exist "%INSTALL_FFMPEG_DIR%" (
+	echo Downloading ffmpeg from %FFMPEG_DOWNLOAD_URL% to %INSTALL_DIR%
+ 	call curl -Lk "%FFMPEG_DOWNLOAD_URL%" > "%INSTALL_DIR%\ffmpeg.zip" || ( echo. && echo ffmpeg failed to download. && goto end )
+	call powershell -command "Expand-Archive -Force '%INSTALL_DIR%\ffmpeg.zip' '%INSTALL_DIR%\'"
+
+	cd "installer_files"
+	setlocal EnableExtensions EnableDelayedExpansion
+
+	for /f "tokens=*" %%f in ('dir /s /b /ad "ffmpeg*"') do (
+		ren "%%f" "ffmpeg"
+	)
+	endlocal
+	setx PATH "%INSTALL_FFMPEG_DIR%\bin\;%PATH%"
+	echo To use videos, you need to restart roop after this installation. 
+	cd ..
+)
+
+@rem check if conda environment was actually created
+if not exist "%INSTALL_ENV_DIR%\python.exe" ( echo. && echo ERROR: Conda environment is empty. && goto end )
+
+@rem activate installer env
+call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniconda hook not found. && goto end )
+
+@rem setup installer env
+echo Launching roop unleashed - please edit windows_run.bat to customize commandline arguments
+call python installer.py %COMMANDLINE_ARGS%
+
+echo.
+echo Done!
+
+:end
+pause
+
+
+

jaa.py

@@ -0,0 +1,355 @@
+"""
+Jaa.py Plugin Framework
+Author: Janvarev Vladislav
+
+Jaa.py - minimalistic one-file plugin framework with no dependencies.
+Main functions:
+- run all plugins files from "plugins" folder, base on filename
+- save each plugin options in "options" folder in JSON text files for further editing
+
+- Plugins
+must located in plugins/ folder
+must have "start(core)" function, that returns manifest dict
+manifest must contain keys "name" and "version"
+can contain "default_options"
+- if contain - options will be saved in "options" folder and reload instead next time
+- if contain - "start_with_options(core,manifest)" function will run with manifest with "options" key
+manifest will be processed in "process_plugin_manifest" function if you override it
+
+- Options (for plugins)
+are saved under "options" folder in JSON format
+created at first run plugin with "default_options"
+updated when plugin change "version"
+
+- Example usage:
+class VoiceAssCore(JaaCore): # class must override JaaCore
+    def __init__(self):
+        JaaCore.__init__(self,__file__)
+  ...
+
+main = VoiceAssCore()
+main.init_plugins(["core"]) # 1 param - first plugins to be initialized
+                            # Good if you need some "core" options/plugin to be loaded before others
+                            # not necessary starts with "plugin_" prefix
+
+also can be run like
+
+main.init_plugins()
+
+- Requirements
+Python 3.5+ (due to dict mix in final_options calc), can be relaxed
+"""
+
+import os
+import traceback
+import json
+
+# here we trying to use termcolor to highlight plugin info and errors during load
+try:
+    from termcolor import cprint
+except Exception as e:
+    # not found? making a stub!
+    def cprint(p,color=None):
+        if color == None:
+            print(p)
+        else:
+            print(str(color).upper(),p)
+
+version = "2.2.0"
+
+class JaaCore:
+    verbose = False
+
+    def __init__(self,root_file = __file__):
+        self.jaaPluginPrefix = "plugin_"
+        self.jaaVersion = version
+        self.jaaRootFolder = os.path.dirname(root_file)
+        self.jaaOptionsPath = self.jaaRootFolder+os.path.sep+"plugin_options"
+        self.jaaShowTracebackOnPluginErrors = False
+        if self.verbose:
+            cprint("JAA.PY v{0} class created!".format(version),"blue")
+
+    # ------------- plugins -----------------
+    def init_plugins(self, list_first_plugins = []):
+        self.plugin_manifests = {}
+
+        # 1. run first plugins first!
+        for modname in list_first_plugins:
+            self.init_plugin(modname)
+
+        # 2. run all plugins from plugins folder
+        from os import listdir
+        from os.path import isfile, join
+        pluginpath = self.jaaRootFolder+"/plugins"
+        files = [f for f in listdir(pluginpath) if isfile(join(pluginpath, f))]
+
+        for fil in files:
+            # print fil[:-3]
+            if fil.startswith(self.jaaPluginPrefix) and fil.endswith(".py"):
+                modfile = fil[:-3]
+                self.init_plugin(modfile)
+
+
+
+    def init_plugin(self,modname):
+        # import
+        try:
+            mod = self.import_plugin("plugins."+modname)
+        except Exception as e:
+            self.print_error("JAA PLUGIN ERROR: {0} error on load: {1}".format(modname, str(e)))
+            return False
+
+        # run start function
+        try:
+            res = mod.start(self)
+        except Exception as e:
+            self.print_error("JAA PLUGIN ERROR: {0} error on start: {1}".format(modname, str(e)))
+            return False
+
+        # if plugin has an options
+        if "default_options" in res:
+            try:
+                # saved options try to read
+                saved_options = {}
+                try:
+                    with open(self.jaaOptionsPath+'/'+modname+'.json', 'r', encoding="utf-8") as f:
+                        s = f.read()
+                    saved_options = json.loads(s)
+                    #print("Saved options", saved_options)
+                except Exception as e:
+                    pass
+
+                res["default_options"]["v"] = res["version"]
+
+
+                # only string needs Python 3.5
+                final_options = {**res["default_options"], **saved_options}
+
+                # if no option found or version is differ from mod version
+                if len(saved_options) == 0 or saved_options["v"] != res["version"]:
+                    final_options["v"] = res["version"]
+                    self.save_plugin_options(modname,final_options)
+
+                res["options"] = final_options
+
+                try:
+                    res2 = mod.start_with_options(self,res)
+                    if res2 != None:
+                        res = res2
+                except Exception as e:
+                    self.print_error("JAA PLUGIN ERROR: {0} error on start_with_options processing: {1}".format(modname, str(e)))
+                    return False
+
+            except Exception as e:
+                self.print_error("JAA PLUGIN ERROR: {0} error on options processing: {1}".format(modname, str(e)))
+                return False
+
+
+        # processing plugin manifest
+        try:
+            # set up name and version
+            plugin_name = res["name"]
+            plugin_version = res["version"]
+
+
+            self.process_plugin_manifest(modname,res)
+
+        except Exception as e:
+            print("JAA PLUGIN ERROR: {0} error on process startup options: {1}".format(modname, str(e)))
+            return False
+
+        self.plugin_manifests[modname] = res
+
+        self.on_succ_plugin_start(modname,plugin_name,plugin_version)
+        return True
+
+    def on_succ_plugin_start(self, modname, plugin_name, plugin_version):
+        if self.verbose:
+            cprint("JAA PLUGIN: {1} {2} ({0}) started!".format(modname, plugin_name, plugin_version))
+
+    def print_error(self,p):
+        cprint(p,"red")
+        if self.jaaShowTracebackOnPluginErrors:
+            traceback.print_exc()
+
+    def import_plugin(self, module_name):
+        import sys
+
+        __import__(module_name)
+
+        if module_name in sys.modules:
+            return sys.modules[module_name]
+        return None
+
+    def save_plugin_options(self,modname,options):
+        # check folder exists
+        if not os.path.exists(self.jaaOptionsPath):
+            os.makedirs(self.jaaOptionsPath)
+
+        str_options = json.dumps(options, sort_keys=True, indent=4, ensure_ascii=False)
+        with open(self.jaaOptionsPath+'/'+modname+'.json', 'w', encoding="utf-8") as f:
+            f.write(str_options)
+            f.close()
+
+    # process manifest must be overrided in inherit class
+    def process_plugin_manifest(self,modname,manifest):
+        print("JAA PLUGIN: {0} manifest dummy procession (override 'process_plugin_manifest' function)".format(modname))
+        return
+
+    def plugin_manifest(self,pluginname):
+        if pluginname in self.plugin_manifests:
+            return self.plugin_manifests[pluginname]
+        return {}
+
+    def plugin_options(self,pluginname):
+        manifest = self.plugin_manifest(pluginname)
+        if "options" in manifest:
+            return manifest["options"]
+        return None
+
+    # ------------ gradio stuff --------------
+    def gradio_save(self,pluginname):
+        print("Saving options for {0}!".format(pluginname))
+        self.save_plugin_options(pluginname,self.plugin_options(pluginname))
+
+    def gradio_upd(self, pluginname, option, val):
+        options = self.plugin_options(pluginname)
+
+        # special case
+        if isinstance(options[option], (list, dict)) and isinstance(val, str):
+            import json
+            try:
+                options[option] = json.loads(val)
+            except Exception as e:
+                print(e)
+                pass
+        else:
+            options[option] = val
+        print(option,val,options)
+
+    def gradio_render_settings_interface(self, title:str="Settings manager", required_fields_to_show_plugin:list=["default_options"]):
+        import gradio as gr
+
+        with gr.Blocks() as gr_interface:
+            gr.Markdown("# {0}".format(title))
+            for pluginname in self.plugin_manifests:
+                manifest = self.plugin_manifests[pluginname]
+
+                # calculate if we show plugin
+                is_show_plugin = False
+                if len(required_fields_to_show_plugin) == 0:
+                    is_show_plugin = True
+                else:
+                    for k in required_fields_to_show_plugin:
+                        if manifest.get(k) is not None:
+                            is_show_plugin = True
+
+                if is_show_plugin:
+                    with gr.Tab(pluginname):
+                        gr.Markdown("## {0} v{1}".format(manifest["name"],manifest["version"]))
+                        if manifest.get("description") is not None:
+                            gr.Markdown(manifest.get("description"))
+
+                        if manifest.get("url") is not None:
+                            gr.Markdown("**URL:** [{0}]({0})".format(manifest.get("url")))
+
+
+                        if "options" in manifest:
+                            options = manifest["options"]
+                            if len(options) > 1: # not only v
+                                text_button = gr.Button("Save options".format(pluginname))
+                                #options_int_list = []
+                                for option in options:
+
+                                    #gr.Label(label=option)
+                                    if option != "v":
+                                        val = options[option]
+                                        label = option
+
+                                        if manifest.get("options_label") is not None:
+                                            if manifest.get("options_label").get(option) is not None:
+                                                label = option+": "+manifest.get("options_label").get(option)
+
+
+                                        if isinstance(val, (bool, )):
+                                            gr_elem = gr.Checkbox(value=val,label=label)
+                                        elif isinstance(val, (dict,list)):
+                                            import json
+                                            gr_elem = gr.Textbox(value=json.dumps(val,ensure_ascii=False), label=label)
+                                        else:
+                                            gr_elem = gr.Textbox(value=val, label=label)
+
+                                        def handler(x,pluginname=pluginname,option=option):
+                                            self.gradio_upd(pluginname, option, x)
+
+                                        gr_elem.change(handler, gr_elem, None)
+
+                                def handler_save(pluginname=pluginname):
+                                    self.gradio_save(pluginname)
+
+                                text_button.click(handler_save,inputs=None,outputs=None)
+                        else:
+                            gr.Markdown("_No options for this plugin_")
+
+        return gr_interface
+
+
+def load_options(options_file=None,py_file=None,default_options={}):
+    # 1. calculating options filename
+    if options_file == None:
+        if py_file == None:
+            raise Exception('JAA: Options or PY file is not defined, cant calc options filename')
+        else:
+            options_file = py_file[:-3]+'.json'
+
+    # 2. try to read saved options
+    saved_options = {}
+    try:
+        with open(options_file, 'r', encoding="utf-8") as f:
+            s = f.read()
+        saved_options = json.loads(s)
+        #print("Saved options", saved_options)
+    except Exception as e:
+        pass
+
+    # 3. calculating final options
+
+    # only string needs Python 3.5
+    final_options = {**default_options, **saved_options}
+
+    # 4. calculating hash from def options to check - is file rewrite needed?
+    import hashlib
+    hash = hashlib.md5((json.dumps(default_options, sort_keys=True)).encode('utf-8')).hexdigest()
+
+    # 5. if no option file found or hash was from other default options
+    if len(saved_options) == 0 or not ("hash" in saved_options.keys()) or saved_options["hash"] != hash:
+        final_options["hash"] = hash
+        #self.save_plugin_options(modname,final_options)
+
+        # saving in file
+        str_options = json.dumps(final_options, sort_keys=True, indent=4, ensure_ascii=False)
+        with open(options_file, 'w', encoding="utf-8") as f:
+            f.write(str_options)
+            f.close()
+
+    return final_options
+
+"""
+The MIT License (MIT)
+Copyright (c) 2021 Janvarev Vladislav
+
+Permission is hereby granted, free of charge, to any person obtaining a copy 
+of this software and associated documentation files (the “Software”), to deal 
+in the Software without restriction, including without limitation the rights to use, 
+copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, 
+and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or 
+substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, 
+INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR 
+PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE 
+FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 
+ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+"""

models/CLIP/rd64-uni-refined.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4956f9a7978a75630b08c9d6ec075b7c51cf43b4751b686e3a011d4012ddc9d
+size 4720707

models/CodeFormer/codeformer.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1009e537e0c2a07d4cabce6355f53cb66767cd4b4297ec7a4a64ca4b8a5684b7
+size 376637898

models/CodeFormer/facelib/detection_Resnet50_Final.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d1de9c2944f2ccddca5f5e010ea5ae64a39845a86311af6fdf30841b0a5a16d
+size 109497761

models/CodeFormer/facelib/parsing_parsenet.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d558d8d0e42c20224f13cf5a29c79eba2d59913419f945545d8cf7b72920de2
+size 85331193

models/CodeFormer/realesrgan/RealESRGAN_x2plus.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49fafd45f8fd7aa8d31ab2a22d14d91b536c34494a5cfe31eb5d89c2fa266abb
+size 67061725

models/DMDNet.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70daeb4b1fd10f241043b587d892a941f2651d7322db02f06ff64b166537f65c
+size 603684323

models/GFPGANv1.4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2cd4703ab14f4d01fd1383a8a8b266f9a5833dacee8e6a79d3bf21a1b6be5ad
+size 348632874

models/inswapper_128.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4a3f08c753cb72d04e10aa0f7dbe3deebbf39567d4ead6dce08e98aa49e16af
+size 554253681

mynewshinyroop/.gitignore

@@ -0,0 +1,2 @@
+# created by virtualenv automatically
+*

mynewshinyroop/Lib/site-packages/_distutils_hack/__init__.py

@@ -0,0 +1,227 @@
+# don't import any costly modules
+import sys
+import os
+
+
+is_pypy = '__pypy__' in sys.builtin_module_names
+
+
+def warn_distutils_present():
+    if 'distutils' not in sys.modules:
+        return
+    if is_pypy and sys.version_info < (3, 7):
+        # PyPy for 3.6 unconditionally imports distutils, so bypass the warning
+        # https://foss.heptapod.net/pypy/pypy/-/blob/be829135bc0d758997b3566062999ee8b23872b4/lib-python/3/site.py#L250
+        return
+    import warnings
+
+    warnings.warn(
+        "Distutils was imported before Setuptools, but importing Setuptools "
+        "also replaces the `distutils` module in `sys.modules`. This may lead "
+        "to undesirable behaviors or errors. To avoid these issues, avoid "
+        "using distutils directly, ensure that setuptools is installed in the "
+        "traditional way (e.g. not an editable install), and/or make sure "
+        "that setuptools is always imported before distutils."
+    )
+
+
+def clear_distutils():
+    if 'distutils' not in sys.modules:
+        return
+    import warnings
+
+    warnings.warn("Setuptools is replacing distutils.")
+    mods = [
+        name
+        for name in sys.modules
+        if name == "distutils" or name.startswith("distutils.")
+    ]
+    for name in mods:
+        del sys.modules[name]
+
+
+def enabled():
+    """
+    Allow selection of distutils by environment variable.
+    """
+    which = os.environ.get('SETUPTOOLS_USE_DISTUTILS', 'local')
+    return which == 'local'
+
+
+def ensure_local_distutils():
+    import importlib
+
+    clear_distutils()
+
+    # With the DistutilsMetaFinder in place,
+    # perform an import to cause distutils to be
+    # loaded from setuptools._distutils. Ref #2906.
+    with shim():
+        importlib.import_module('distutils')
+
+    # check that submodules load as expected
+    core = importlib.import_module('distutils.core')
+    assert '_distutils' in core.__file__, core.__file__
+    assert 'setuptools._distutils.log' not in sys.modules
+
+
+def do_override():
+    """
+    Ensure that the local copy of distutils is preferred over stdlib.
+
+    See https://github.com/pypa/setuptools/issues/417#issuecomment-392298401
+    for more motivation.
+    """
+    if enabled():
+        warn_distutils_present()
+        ensure_local_distutils()
+
+
+class _TrivialRe:
+    def __init__(self, *patterns):
+        self._patterns = patterns
+
+    def match(self, string):
+        return all(pat in string for pat in self._patterns)
+
+
+class DistutilsMetaFinder:
+    def find_spec(self, fullname, path, target=None):
+        # optimization: only consider top level modules and those
+        # found in the CPython test suite.
+        if path is not None and not fullname.startswith('test.'):
+            return
+
+        method_name = 'spec_for_{fullname}'.format(**locals())
+        method = getattr(self, method_name, lambda: None)
+        return method()
+
+    def spec_for_distutils(self):
+        if self.is_cpython():
+            return
+
+        import importlib
+        import importlib.abc
+        import importlib.util
+
+        try:
+            mod = importlib.import_module('setuptools._distutils')
+        except Exception:
+            # There are a couple of cases where setuptools._distutils
+            # may not be present:
+            # - An older Setuptools without a local distutils is
+            #   taking precedence. Ref #2957.
+            # - Path manipulation during sitecustomize removes
+            #   setuptools from the path but only after the hook
+            #   has been loaded. Ref #2980.
+            # In either case, fall back to stdlib behavior.
+            return
+
+        class DistutilsLoader(importlib.abc.Loader):
+            def create_module(self, spec):
+                mod.__name__ = 'distutils'
+                return mod
+
+            def exec_module(self, module):
+                pass
+
+        return importlib.util.spec_from_loader(
+            'distutils', DistutilsLoader(), origin=mod.__file__
+        )
+
+    @staticmethod
+    def is_cpython():
+        """
+        Suppress supplying distutils for CPython (build and tests).
+        Ref #2965 and #3007.
+        """
+        return os.path.isfile('pybuilddir.txt')
+
+    def spec_for_pip(self):
+        """
+        Ensure stdlib distutils when running under pip.
+        See pypa/pip#8761 for rationale.
+        """
+        if sys.version_info >= (3, 12) or self.pip_imported_during_build():
+            return
+        clear_distutils()
+        self.spec_for_distutils = lambda: None
+
+    @classmethod
+    def pip_imported_during_build(cls):
+        """
+        Detect if pip is being imported in a build script. Ref #2355.
+        """
+        import traceback
+
+        return any(
+            cls.frame_file_is_setup(frame) for frame, line in traceback.walk_stack(None)
+        )
+
+    @staticmethod
+    def frame_file_is_setup(frame):
+        """
+        Return True if the indicated frame suggests a setup.py file.
+        """
+        # some frames may not have __file__ (#2940)
+        return frame.f_globals.get('__file__', '').endswith('setup.py')
+
+    def spec_for_sensitive_tests(self):
+        """
+        Ensure stdlib distutils when running select tests under CPython.
+
+        python/cpython#91169
+        """
+        clear_distutils()
+        self.spec_for_distutils = lambda: None
+
+    sensitive_tests = (
+        [
+            'test.test_distutils',
+            'test.test_peg_generator',
+            'test.test_importlib',
+        ]
+        if sys.version_info < (3, 10)
+        else [
+            'test.test_distutils',
+        ]
+    )
+
+
+for name in DistutilsMetaFinder.sensitive_tests:
+    setattr(
+        DistutilsMetaFinder,
+        f'spec_for_{name}',
+        DistutilsMetaFinder.spec_for_sensitive_tests,
+    )
+
+
+DISTUTILS_FINDER = DistutilsMetaFinder()
+
+
+def add_shim():
+    DISTUTILS_FINDER in sys.meta_path or insert_shim()
+
+
+class shim:
+    def __enter__(self):
+        insert_shim()
+
+    def __exit__(self, exc, value, tb):
+        _remove_shim()
+
+
+def insert_shim():
+    sys.meta_path.insert(0, DISTUTILS_FINDER)
+
+
+def _remove_shim():
+    try:
+        sys.meta_path.remove(DISTUTILS_FINDER)
+    except ValueError:
+        pass
+
+
+if sys.version_info < (3, 12):
+    # DistutilsMetaFinder can only be disabled in Python < 3.12 (PEP 632)
+    remove_shim = _remove_shim

mynewshinyroop/Lib/site-packages/_distutils_hack/override.py

@@ -0,0 +1 @@
+__import__('_distutils_hack').do_override()

mynewshinyroop/Lib/site-packages/_virtualenv.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69ac3d8f27e679c81b94ab30b3b56e9cd138219b1ba94a1fa3606d5a76a1433d
+size 18

mynewshinyroop/Lib/site-packages/_virtualenv.py

@@ -0,0 +1,102 @@
+"""Patches that are applied at runtime to the virtual environment."""
+
+from __future__ import annotations
+
+import os
+import sys
+from contextlib import suppress
+
+VIRTUALENV_PATCH_FILE = os.path.join(__file__)
+
+
+def patch_dist(dist):
+    """
+    Distutils allows user to configure some arguments via a configuration file:
+    https://docs.python.org/3/install/index.html#distutils-configuration-files.
+
+    Some of this arguments though don't make sense in context of the virtual environment files, let's fix them up.
+    """  # noqa: D205
+    # we cannot allow some install config as that would get packages installed outside of the virtual environment
+    old_parse_config_files = dist.Distribution.parse_config_files
+
+    def parse_config_files(self, *args, **kwargs):
+        result = old_parse_config_files(self, *args, **kwargs)
+        install = self.get_option_dict("install")
+
+        if "prefix" in install:  # the prefix governs where to install the libraries
+            install["prefix"] = VIRTUALENV_PATCH_FILE, os.path.abspath(sys.prefix)
+        for base in ("purelib", "platlib", "headers", "scripts", "data"):
+            key = f"install_{base}"
+            if key in install:  # do not allow global configs to hijack venv paths
+                install.pop(key, None)
+        return result
+
+    dist.Distribution.parse_config_files = parse_config_files
+
+
+# Import hook that patches some modules to ignore configuration values that break package installation in case
+# of virtual environments.
+_DISTUTILS_PATCH = "distutils.dist", "setuptools.dist"
+# https://docs.python.org/3/library/importlib.html#setting-up-an-importer
+
+
+class _Finder:
+    """A meta path finder that allows patching the imported distutils modules."""
+
+    fullname = None
+
+    # lock[0] is threading.Lock(), but initialized lazily to avoid importing threading very early at startup,
+    # because there are gevent-based applications that need to be first to import threading by themselves.
+    # See https://github.com/pypa/virtualenv/issues/1895 for details.
+    lock = []  # noqa: RUF012
+
+    def find_spec(self, fullname, path, target=None):  # noqa: ARG002
+        if fullname in _DISTUTILS_PATCH and self.fullname is None:
+            # initialize lock[0] lazily
+            if len(self.lock) == 0:
+                import threading
+
+                lock = threading.Lock()
+                # there is possibility that two threads T1 and T2 are simultaneously running into find_spec,
+                # observing .lock as empty, and further going into hereby initialization. However due to the GIL,
+                # list.append() operation is atomic and this way only one of the threads will "win" to put the lock
+                # - that every thread will use - into .lock[0].
+                # https://docs.python.org/3/faq/library.html#what-kinds-of-global-value-mutation-are-thread-safe
+                self.lock.append(lock)
+
+            from functools import partial
+            from importlib.util import find_spec
+
+            with self.lock[0]:
+                self.fullname = fullname
+                try:
+                    spec = find_spec(fullname, path)
+                    if spec is not None:
+                        # https://www.python.org/dev/peps/pep-0451/#how-loading-will-work
+                        is_new_api = hasattr(spec.loader, "exec_module")
+                        func_name = "exec_module" if is_new_api else "load_module"
+                        old = getattr(spec.loader, func_name)
+                        func = self.exec_module if is_new_api else self.load_module
+                        if old is not func:
+                            with suppress(AttributeError):  # C-Extension loaders are r/o such as zipimporter with <3.7
+                                setattr(spec.loader, func_name, partial(func, old))
+                        return spec
+                finally:
+                    self.fullname = None
+        return None
+
+    @staticmethod
+    def exec_module(old, module):
+        old(module)
+        if module.__name__ in _DISTUTILS_PATCH:
+            patch_dist(module)
+
+    @staticmethod
+    def load_module(old, name):
+        module = old(name)
+        if module.__name__ in _DISTUTILS_PATCH:
+            patch_dist(module)
+        return module
+
+
+sys.meta_path.insert(0, _Finder())