Spaces:
Sleeping
Sleeping
Added support to download prediction
Browse files- demo/src/gui.py +40 -15
- demo/src/inference.py +5 -1
demo/src/gui.py
CHANGED
@@ -59,7 +59,8 @@ class WebUI:
|
|
59 |
visible=True,
|
60 |
elem_id="model-3d",
|
61 |
camera_position=[90, 180, 768],
|
62 |
-
|
|
|
63 |
|
64 |
def set_class_name(self, value):
|
65 |
LOGGER.info(f"Changed task to: {value}")
|
@@ -75,22 +76,31 @@ class WebUI:
|
|
75 |
|
76 |
def process(self, mesh_file_name):
|
77 |
path = mesh_file_name.name
|
|
|
|
|
|
|
78 |
run_model(
|
79 |
path,
|
80 |
model_path=os.path.join(self.cwd, "resources/models/"),
|
81 |
task=self.class_names[self.class_name],
|
82 |
name=self.result_names[self.class_name],
|
|
|
83 |
)
|
84 |
LOGGER.info("Converting prediction NIfTI to OBJ...")
|
85 |
-
nifti_to_obj("
|
86 |
|
87 |
LOGGER.info("Loading CT to numpy...")
|
88 |
self.images = load_ct_to_numpy(path)
|
89 |
|
90 |
LOGGER.info("Loading prediction volume to numpy..")
|
91 |
-
self.pred_images = load_pred_volume_to_numpy("
|
92 |
|
93 |
return "./prediction.obj"
|
|
|
|
|
|
|
|
|
|
|
94 |
|
95 |
def get_img_pred_pair(self, k):
|
96 |
k = int(k)
|
@@ -98,7 +108,6 @@ class WebUI:
|
|
98 |
self.combine_ct_and_seg(self.images[k], self.pred_images[k]),
|
99 |
visible=True,
|
100 |
elem_id="model-2d",
|
101 |
-
).style(
|
102 |
color_map={self.class_name: "#ffae00"},
|
103 |
height=512,
|
104 |
width=512,
|
@@ -122,9 +131,7 @@ class WebUI:
|
|
122 |
autoscroll=True,
|
123 |
elem_id="logs",
|
124 |
show_copy_button=True,
|
125 |
-
scroll_to_output=False,
|
126 |
container=True,
|
127 |
-
line_breaks=True,
|
128 |
)
|
129 |
demo.load(read_logs, None, logs, every=1)
|
130 |
|
@@ -160,7 +167,7 @@ class WebUI:
|
|
160 |
label="Task",
|
161 |
info="Which structure to segment.",
|
162 |
multiselect=False,
|
163 |
-
|
164 |
)
|
165 |
model_selector.input(
|
166 |
fn=lambda x: self.set_class_name(x),
|
@@ -173,16 +180,33 @@ class WebUI:
|
|
173 |
"Run analysis",
|
174 |
variant="primary",
|
175 |
elem_id="run-button",
|
176 |
-
|
177 |
-
|
178 |
size="lg",
|
179 |
)
|
|
|
|
|
|
|
|
|
180 |
run_btn.click(
|
181 |
fn=lambda x: self.process(x),
|
182 |
inputs=file_output,
|
183 |
outputs=self.volume_renderer,
|
184 |
)
|
185 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
186 |
with gr.Row():
|
187 |
gr.Examples(
|
188 |
examples=[
|
@@ -202,16 +226,17 @@ class WebUI:
|
|
202 |
)
|
203 |
|
204 |
with gr.Row():
|
205 |
-
with gr.Box():
|
206 |
with gr.Column():
|
207 |
# create dummy image to be replaced by loaded images
|
208 |
t = gr.AnnotatedImage(
|
209 |
visible=True, elem_id="model-2d"
|
210 |
-
).style(
|
211 |
-
color_map={self.class_name: "#ffae00"},
|
212 |
-
height=512,
|
213 |
-
width=512,
|
214 |
)
|
|
|
|
|
|
|
|
|
|
|
215 |
|
216 |
self.slider.input(
|
217 |
self.get_img_pred_pair,
|
@@ -221,7 +246,7 @@ class WebUI:
|
|
221 |
|
222 |
self.slider.render()
|
223 |
|
224 |
-
with gr.Box():
|
225 |
self.volume_renderer.render()
|
226 |
|
227 |
# sharing app publicly -> share=True:
|
|
|
59 |
visible=True,
|
60 |
elem_id="model-3d",
|
61 |
camera_position=[90, 180, 768],
|
62 |
+
height=512,
|
63 |
+
)
|
64 |
|
65 |
def set_class_name(self, value):
|
66 |
LOGGER.info(f"Changed task to: {value}")
|
|
|
76 |
|
77 |
def process(self, mesh_file_name):
|
78 |
path = mesh_file_name.name
|
79 |
+
curr = path.split("/")[-1]
|
80 |
+
self.extension = ".".join(curr.split(".")[1:])
|
81 |
+
self.filename = curr.split(".")[0] + "-" + self.class_names[self.class_name]
|
82 |
run_model(
|
83 |
path,
|
84 |
model_path=os.path.join(self.cwd, "resources/models/"),
|
85 |
task=self.class_names[self.class_name],
|
86 |
name=self.result_names[self.class_name],
|
87 |
+
output_filename=self.filename + "." + self.extension
|
88 |
)
|
89 |
LOGGER.info("Converting prediction NIfTI to OBJ...")
|
90 |
+
nifti_to_obj(path=self.filename + "." + self.extension)
|
91 |
|
92 |
LOGGER.info("Loading CT to numpy...")
|
93 |
self.images = load_ct_to_numpy(path)
|
94 |
|
95 |
LOGGER.info("Loading prediction volume to numpy..")
|
96 |
+
self.pred_images = load_pred_volume_to_numpy(self.filename + "." + self.extension)
|
97 |
|
98 |
return "./prediction.obj"
|
99 |
+
|
100 |
+
def download_prediction(self):
|
101 |
+
if (not self.filename) or (not self.extension):
|
102 |
+
LOGGER.error("The prediction is not available or ready to download. Wait until the result is available in the 3D viewer.")
|
103 |
+
return self.filename + "." + self.extension
|
104 |
|
105 |
def get_img_pred_pair(self, k):
|
106 |
k = int(k)
|
|
|
108 |
self.combine_ct_and_seg(self.images[k], self.pred_images[k]),
|
109 |
visible=True,
|
110 |
elem_id="model-2d",
|
|
|
111 |
color_map={self.class_name: "#ffae00"},
|
112 |
height=512,
|
113 |
width=512,
|
|
|
131 |
autoscroll=True,
|
132 |
elem_id="logs",
|
133 |
show_copy_button=True,
|
|
|
134 |
container=True,
|
|
|
135 |
)
|
136 |
demo.load(read_logs, None, logs, every=1)
|
137 |
|
|
|
167 |
label="Task",
|
168 |
info="Which structure to segment.",
|
169 |
multiselect=False,
|
170 |
+
scale=1.0,
|
171 |
)
|
172 |
model_selector.input(
|
173 |
fn=lambda x: self.set_class_name(x),
|
|
|
180 |
"Run analysis",
|
181 |
variant="primary",
|
182 |
elem_id="run-button",
|
183 |
+
#scale=1.0,
|
184 |
+
# size=1.0,
|
185 |
size="lg",
|
186 |
)
|
187 |
+
#.style(
|
188 |
+
# full_width=False,
|
189 |
+
# size="lg",
|
190 |
+
#)
|
191 |
run_btn.click(
|
192 |
fn=lambda x: self.process(x),
|
193 |
inputs=file_output,
|
194 |
outputs=self.volume_renderer,
|
195 |
)
|
196 |
|
197 |
+
download_btn = gr.DownloadButton(
|
198 |
+
"Download the result as NIfTI",
|
199 |
+
visible=True,
|
200 |
+
variant="secondary",
|
201 |
+
# scale=1.0,
|
202 |
+
size="sm",
|
203 |
+
)
|
204 |
+
download_btn.click(
|
205 |
+
fn=self.download_prediction,
|
206 |
+
inputs=None,
|
207 |
+
outputs=download_btn,
|
208 |
+
)
|
209 |
+
|
210 |
with gr.Row():
|
211 |
gr.Examples(
|
212 |
examples=[
|
|
|
226 |
)
|
227 |
|
228 |
with gr.Row():
|
229 |
+
with gr.Group(): #gr.Box():
|
230 |
with gr.Column():
|
231 |
# create dummy image to be replaced by loaded images
|
232 |
t = gr.AnnotatedImage(
|
233 |
visible=True, elem_id="model-2d"
|
|
|
|
|
|
|
|
|
234 |
)
|
235 |
+
#.style(
|
236 |
+
# color_map={self.class_name: "#ffae00"},
|
237 |
+
# height=512,
|
238 |
+
# width=512,
|
239 |
+
#)
|
240 |
|
241 |
self.slider.input(
|
242 |
self.get_img_pred_pair,
|
|
|
246 |
|
247 |
self.slider.render()
|
248 |
|
249 |
+
with gr.Group(): #gr.Box():
|
250 |
self.volume_renderer.render()
|
251 |
|
252 |
# sharing app publicly -> share=True:
|
demo/src/inference.py
CHANGED
@@ -11,6 +11,7 @@ def run_model(
|
|
11 |
verbose: str = "info",
|
12 |
task: str = "CT_Airways",
|
13 |
name: str = "Airways",
|
|
|
14 |
):
|
15 |
if verbose == "debug":
|
16 |
logging.getLogger().setLevel(logging.DEBUG)
|
@@ -27,6 +28,9 @@ def run_model(
|
|
27 |
if os.path.exists("./result/"):
|
28 |
shutil.rmtree("./result/")
|
29 |
|
|
|
|
|
|
|
30 |
patient_directory = ""
|
31 |
output_path = ""
|
32 |
try:
|
@@ -84,7 +88,7 @@ def run_model(
|
|
84 |
+ "-t1gd_annotation-"
|
85 |
+ name
|
86 |
+ ".nii.gz",
|
87 |
-
|
88 |
)
|
89 |
# Clean-up
|
90 |
if os.path.exists(patient_directory):
|
|
|
11 |
verbose: str = "info",
|
12 |
task: str = "CT_Airways",
|
13 |
name: str = "Airways",
|
14 |
+
output_filename: str = None,
|
15 |
):
|
16 |
if verbose == "debug":
|
17 |
logging.getLogger().setLevel(logging.DEBUG)
|
|
|
28 |
if os.path.exists("./result/"):
|
29 |
shutil.rmtree("./result/")
|
30 |
|
31 |
+
if output_filename is None:
|
32 |
+
raise ValueError("Please, set output_filename.")
|
33 |
+
|
34 |
patient_directory = ""
|
35 |
output_path = ""
|
36 |
try:
|
|
|
88 |
+ "-t1gd_annotation-"
|
89 |
+ name
|
90 |
+ ".nii.gz",
|
91 |
+
output_filename,
|
92 |
)
|
93 |
# Clean-up
|
94 |
if os.path.exists(patient_directory):
|