Hugging Face's logo

Spaces:
facebook
/
vggt
Running on Zero

App Files Community
JianyuanWang commited on
Commit
d35072c
·
1 Parent(s): ae06e9d

update

Browse files
Files changed (14) hide show
  1. app.py +3 -3
  2. examples/room/images/{no_overlap_2.HEIC.JPG → no_overlap_2.jpg} +2 -2
  3. examples/room/images/{no_overlap_3.HEIC.JPG → no_overlap_3.jpg} +2 -2
  4. examples/room/images/{no_overlap_4.HEIC.JPG → no_overlap_4.jpg} +2 -2
  5. examples/room/images/{no_overlap_5.HEIC.JPG → no_overlap_5.jpg} +2 -2
  6. examples/room/images/no_overlap_6.HEIC.JPG +0 -3
  7. examples/room/images/no_overlap_6.jpg +3 -0
  8. examples/room/images/no_overlap_7.HEIC.JPG +0 -3
  9. examples/room/images/no_overlap_7.jpg +3 -0
  10. examples/room/images/no_overlap_8.HEIC.JPG +0 -3
  11. examples/room/images/no_overlap_8.jpg +3 -0
  12. examples/single/images/model_was_never_trained_on_single_image.jpg +3 -0
  13. examples/videos/single.mp4 +3 -0
  14. gradio_util.py +5 -0
app.py CHANGED
@@ -256,15 +256,15 @@ with gr.Blocks(css="""
256
  # 🏛️ VGGT: Visual Geometry Grounded Transformer
257
 
258
  <div style="font-size: 16px; line-height: 1.5;">
259
- <p><strong>Alpha version</strong> (under active development)</p>
260
 
261
  <p>Upload a video or images to create a 3D reconstruction. Once your media appears in the left panel, click the "Reconstruct" button to begin processing.</p>
262
 
263
  <h3>Usage Tips:</h3>
264
  <ol>
265
- <li>After reconstruction, you can fine-tune the visualization by adjusting the confidence threshold or selecting specific frames to display, then click "Update Visualization".</li>
266
  <li>Performance note: While the model itself processes quickly (~0.2 seconds), initial setup and visualization may take longer. First-time use requires downloading model weights, and rendering dense point clouds can be resource-intensive.</li>
267
- <li>Known limitation: The model currently exhibits inconsistent behavior with videos centered around human subjects. This issue is being addressed in upcoming updates.</li>
268
  </ol>
269
  </div>
270
  """)
 
256
  # 🏛️ VGGT: Visual Geometry Grounded Transformer
257
 
258
  <div style="font-size: 16px; line-height: 1.5;">
259
+ <p><strong>Alpha version</strong> (under active change)</p>
260
 
261
  <p>Upload a video or images to create a 3D reconstruction. Once your media appears in the left panel, click the "Reconstruct" button to begin processing.</p>
262
 
263
  <h3>Usage Tips:</h3>
264
  <ol>
265
+ <li>After reconstruction, you can adjust the visualization by adjusting the confidence threshold, selecting specific frames to display, and so on.</li>
266
  <li>Performance note: While the model itself processes quickly (~0.2 seconds), initial setup and visualization may take longer. First-time use requires downloading model weights, and rendering dense point clouds can be resource-intensive.</li>
267
+ <li>Known limitation: The model currently exhibits weird behavior with videos centered around human subjects. This issue is being addressed in upcoming updates.</li>
268
  </ol>
269
  </div>
270
  """)
examples/room/images/{no_overlap_2.HEIC.JPG → no_overlap_2.jpg} RENAMED
File without changes
examples/room/images/{no_overlap_3.HEIC.JPG → no_overlap_3.jpg} RENAMED
File without changes
examples/room/images/{no_overlap_4.HEIC.JPG → no_overlap_4.jpg} RENAMED
File without changes
examples/room/images/{no_overlap_5.HEIC.JPG → no_overlap_5.jpg} RENAMED
File without changes
examples/room/images/no_overlap_6.HEIC.JPG DELETED

Git LFS Details

  • SHA256: ef867329251646dd1a9a65e53d520abe1ae312f4b03020def00be1d82e079119
  • Pointer size: 131 Bytes
  • Size of remote file: 255 kB
examples/room/images/no_overlap_6.jpg ADDED

Git LFS Details

  • SHA256: 01cf8937c2da430ef49e9c0cb23a8031a698eebf2dd1261a37a5c1ee28f5a7f5
  • Pointer size: 131 Bytes
  • Size of remote file: 271 kB
examples/room/images/no_overlap_7.HEIC.JPG DELETED

Git LFS Details

  • SHA256: e1f24a0d124f2a884a8f98a887f6d4c99ac8ed6251f25cd1e59a8ab287328d85
  • Pointer size: 131 Bytes
  • Size of remote file: 256 kB
examples/room/images/no_overlap_7.jpg ADDED

Git LFS Details

  • SHA256: 189a30e8bd6445c972eb6a8c31581e9af6d0bbc03b0345fba5ca023e678f5492
  • Pointer size: 131 Bytes
  • Size of remote file: 261 kB
examples/room/images/no_overlap_8.HEIC.JPG DELETED

Git LFS Details

  • SHA256: 127a830eb92e5a33ea4f01002ef12073e827789e47f4fb59910fa050793eba79
  • Pointer size: 131 Bytes
  • Size of remote file: 128 kB
examples/room/images/no_overlap_8.jpg ADDED

Git LFS Details

  • SHA256: db13926aab6bcc0f7c4903839ccb0dc554ab3276fdcf73e0f304b596f5c15221
  • Pointer size: 131 Bytes
  • Size of remote file: 191 kB
examples/single/images/model_was_never_trained_on_single_image.jpg ADDED

Git LFS Details

  • SHA256: bf60ce3879f627e8886280cc61442174c91908894a5b059681341fed600f7db3
  • Pointer size: 131 Bytes
  • Size of remote file: 769 kB
examples/videos/single.mp4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f7d5b07fbceb13043c4e8350d685150fd92bc6efec9ff959cef3aa11cf69f86e
3
+ size 909011
gradio_util.py CHANGED
@@ -108,6 +108,11 @@ def demo_predictions_to_glb(predictions, conf_thres=3.0, filter_by_frames="all",
108
  colors_rgb = colors_rgb[conf_mask]
109
 
110
 
 
 
 
 
 
111
 
112
  if vertices_3d is None or np.asarray(vertices_3d).size == 0:
113
  vertices_3d = np.array([[1, 0, 0]])
 
108
  colors_rgb = colors_rgb[conf_mask]
109
 
110
 
111
+
112
+ # resize_scale = 0.25
113
+ # vertices_3d = vertices_3d * resize_scale
114
+ # camera_matrices[:, :3, 3] = camera_matrices[:, :3, 3] * resize_scale
115
+
116
 
117
  if vertices_3d is None or np.asarray(vertices_3d).size == 0:
118
  vertices_3d = np.array([[1, 0, 0]])