Update app.py
Browse files
app.py
CHANGED
@@ -188,132 +188,132 @@ with gr.Blocks() as demo:
|
|
188 |
with gr.Row():
|
189 |
gr.Markdown("""# D-Edit""")
|
190 |
|
191 |
-
if 1:
|
192 |
-
with gr.Row():
|
193 |
-
with gr.Column():
|
194 |
-
canvas = gr.Image(value = None, type="numpy", label="Show Mask", show_label=True, height=LENGTH, width=LENGTH, interactive=True)
|
195 |
-
example_inps = [['./img.png'],['./img2.png'],['./img3.png'],['./img4.png']]
|
196 |
-
gr.Examples(examples=example_inps, inputs=[canvas],
|
197 |
-
label='examples', cache_examples='lazy', outputs=[],
|
198 |
-
fn=change_image)
|
199 |
-
gr.Markdown(f"Each image must first undergo segmentation. Afterwards, you can modify the \n mask ID and the prompt for image editing, then proceed with the editing process. \n The link of D-edit paper: [https://arxiv.org/abs/2403.04880v2](https://arxiv.org/abs/2403.04880v2), [https://huggingface.co/papers/2403.04880](https://huggingface.co/papers/2403.04880)")
|
200 |
|
201 |
-
|
202 |
-
|
203 |
-
|
204 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
205 |
|
206 |
-
|
207 |
-
|
208 |
-
gr.Markdown("""<p style="text-align: center; font-size: 20px">Edit Mask (Do not change it during the editing process)</p>""")
|
209 |
-
slider = gr.Slider(0, 20, step=1, label = 'mask id', visible=False)
|
210 |
-
label = gr.Text(label='label')
|
211 |
|
|
|
212 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
213 |
|
214 |
-
|
215 |
-
|
216 |
|
217 |
-
|
218 |
-
gr.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
219 |
with gr.Accordion(label="Advanced settings", open=False):
|
220 |
-
|
221 |
-
|
222 |
-
|
223 |
-
|
224 |
-
|
225 |
-
diffusion_model_learning_rate = gr.Textbox(value="0.0002", label="UNet Optimization: Learning rate", interactive= True )
|
226 |
-
max_diffusion_train_steps = gr.Number(value="28", label="UNet Optimization: Learning rate: Training steps", interactive= True )
|
227 |
-
|
228 |
-
train_batch_size = gr.Number(value="20", label="Batch size", interactive= True )
|
229 |
-
gradient_accumulation_steps=gr.Number(value="2", label="Gradient accumulation", interactive= True )
|
230 |
|
231 |
-
|
|
|
232 |
mask_np_list,
|
233 |
mask_label_list,
|
234 |
image,
|
235 |
-
opt_flag,
|
236 |
num_tokens,
|
237 |
-
|
238 |
-
|
239 |
-
|
240 |
-
|
241 |
-
|
242 |
-
|
243 |
):
|
244 |
-
|
245 |
-
|
246 |
-
|
247 |
-
|
248 |
-
|
249 |
-
|
250 |
-
|
251 |
-
|
252 |
-
|
253 |
-
|
254 |
-
|
255 |
-
|
256 |
-
|
257 |
-
|
258 |
-
|
259 |
-
|
260 |
-
|
261 |
-
|
262 |
-
|
263 |
-
|
264 |
-
return "Error: use a smaller batch size or try latter."#,gr.Button("Step 3. Run Editing",interactive = False)
|
265 |
-
|
266 |
-
|
267 |
-
|
268 |
-
|
269 |
-
with gr.Row():
|
270 |
-
with gr.Column():
|
271 |
-
canvas_text_edit = gr.Image(value = None, type = "pil", label="Editing results", show_label=True,visible = True)
|
272 |
-
# canvas_text_edit = gr.Gallery(label = "Edited results")
|
273 |
-
|
274 |
-
with gr.Column():
|
275 |
-
gr.Markdown("""<p style="text-align: center; font-size: 20px">Editing setting</p>""")
|
276 |
-
tgt_prompt = gr.Textbox(value="text prompt", label="Editing: Text prompt", interactive= True )
|
277 |
-
with gr.Accordion(label="Advanced settings", open=False):
|
278 |
-
slider2 = gr.Slider(0, 20, step=1, label = 'mask id', visible=False)
|
279 |
-
guidance_scale = gr.Textbox(value="5", label="Editing: CFG guidance scale", interactive= True )
|
280 |
-
num_sampling_steps = gr.Number(value="20", label="Editing: Sampling steps", interactive= True )
|
281 |
-
edge_thickness = gr.Number(value="10", label="Editing: Edge thickness", interactive= True )
|
282 |
-
strength = gr.Textbox(value="0.5", label="Editing: Mask strength", interactive= True )
|
283 |
-
|
284 |
-
add_button = gr.Button("Step 2. Run Editing",interactive = True)
|
285 |
-
def run_edit_text_wrapper(
|
286 |
-
mask_np_list,
|
287 |
-
mask_label_list,
|
288 |
-
image,
|
289 |
-
num_tokens,
|
290 |
-
guidance_scale,
|
291 |
-
num_sampling_steps ,
|
292 |
-
strength ,
|
293 |
-
edge_thickness,
|
294 |
-
tgt_prompt ,
|
295 |
-
tgt_index
|
296 |
-
):
|
297 |
-
|
298 |
-
run_edit_text = partial(
|
299 |
-
run_main,
|
300 |
-
mask_np_list=mask_np_list,
|
301 |
-
mask_label_list=mask_label_list,
|
302 |
-
image_gt=np.array(image),
|
303 |
-
load_trained=True,
|
304 |
-
text=True,
|
305 |
-
num_tokens = int(num_tokens_global.value),
|
306 |
-
guidance_scale = float(guidance_scale),
|
307 |
-
num_sampling_steps = int(num_sampling_steps),
|
308 |
-
strength = float(strength),
|
309 |
-
edge_thickness = int(edge_thickness),
|
310 |
-
num_imgs = 1,
|
311 |
-
tgt_prompt = tgt_prompt,
|
312 |
-
tgt_index = int(tgt_index)
|
313 |
-
)
|
314 |
-
run_edit_text()
|
315 |
-
gr.Info('Image editing completed.')
|
316 |
-
return load_pil_img()
|
317 |
|
318 |
|
319 |
|
|
|
188 |
with gr.Row():
|
189 |
gr.Markdown("""# D-Edit""")
|
190 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
191 |
|
192 |
+
with gr.Row():
|
193 |
+
with gr.Column():
|
194 |
+
canvas = gr.Image(value = None, type="numpy", label="Show Mask", show_label=True, height=LENGTH, width=LENGTH, interactive=True)
|
195 |
+
example_inps = [['./img.png'],['./img2.png'],['./img3.png'],['./img4.png']]
|
196 |
+
gr.Examples(examples=example_inps, inputs=[canvas],
|
197 |
+
label='examples', cache_examples='lazy', outputs=[],
|
198 |
+
fn=change_image)
|
199 |
+
gr.Markdown(f"Each image must first undergo segmentation. Afterwards, you can modify the \n mask ID and the prompt for image editing, then proceed with the editing process. \n The link of D-edit paper: [https://arxiv.org/abs/2403.04880v2](https://arxiv.org/abs/2403.04880v2), [https://huggingface.co/papers/2403.04880](https://huggingface.co/papers/2403.04880)")
|
200 |
+
|
201 |
+
with gr.Column():
|
202 |
+
result_info0 = gr.Text(label="Response")
|
203 |
+
segment_button = gr.Button("Step 1. Run segmentation")
|
204 |
+
flag = gr.State(False)
|
205 |
+
|
206 |
+
# mask_np_list_updated.value = copy.deepcopy(mask_np_list.value) #!!
|
207 |
+
mask_np_list_updated = mask_np_list
|
208 |
+
gr.Markdown("""<p style="text-align: center; font-size: 20px">Edit Mask (Do not change it during the editing process)</p>""")
|
209 |
+
slider = gr.Slider(0, 20, step=1, label = 'mask id', visible=False)
|
210 |
+
label = gr.Text(label='label')
|
211 |
|
212 |
+
|
213 |
+
|
|
|
|
|
|
|
214 |
|
215 |
+
result_info = gr.Text(label="Response")
|
216 |
|
217 |
+
opt_flag = gr.State(0)
|
218 |
+
gr.Markdown("""<p style="text-align: center; font-size: 20px">Optimization settings</p>""")
|
219 |
+
with gr.Accordion(label="Advanced settings", open=False):
|
220 |
+
num_tokens = gr.Number(value="5", label="num tokens to represent each object", interactive= True)
|
221 |
+
num_tokens_global = num_tokens
|
222 |
+
embedding_learning_rate = gr.Textbox(value="0.00025", label="Embedding optimization: Learning rate", interactive= True )
|
223 |
+
max_emb_train_steps = gr.Number(value="6", label="embedding optimization: Training steps", interactive= True )
|
224 |
|
225 |
+
diffusion_model_learning_rate = gr.Textbox(value="0.0002", label="UNet Optimization: Learning rate", interactive= True )
|
226 |
+
max_diffusion_train_steps = gr.Number(value="28", label="UNet Optimization: Learning rate: Training steps", interactive= True )
|
227 |
|
228 |
+
train_batch_size = gr.Number(value="20", label="Batch size", interactive= True )
|
229 |
+
gradient_accumulation_steps=gr.Number(value="2", label="Gradient accumulation", interactive= True )
|
230 |
+
|
231 |
+
def run_optimization_wrapper (
|
232 |
+
mask_np_list,
|
233 |
+
mask_label_list,
|
234 |
+
image,
|
235 |
+
opt_flag,
|
236 |
+
num_tokens,
|
237 |
+
embedding_learning_rate ,
|
238 |
+
max_emb_train_steps ,
|
239 |
+
diffusion_model_learning_rate ,
|
240 |
+
max_diffusion_train_steps,
|
241 |
+
train_batch_size,
|
242 |
+
gradient_accumulation_steps,
|
243 |
+
):
|
244 |
+
try:
|
245 |
+
run_optimization = partial(
|
246 |
+
run_main,
|
247 |
+
mask_np_list=mask_np_list,
|
248 |
+
mask_label_list=mask_label_list,
|
249 |
+
image_gt=np.array(image),
|
250 |
+
num_tokens=int(num_tokens),
|
251 |
+
embedding_learning_rate = float(embedding_learning_rate),
|
252 |
+
max_emb_train_steps = int(max_emb_train_steps),
|
253 |
+
diffusion_model_learning_rate= float(diffusion_model_learning_rate),
|
254 |
+
max_diffusion_train_steps = int(max_diffusion_train_steps),
|
255 |
+
train_batch_size=int(train_batch_size),
|
256 |
+
gradient_accumulation_steps=int(gradient_accumulation_steps)
|
257 |
+
)
|
258 |
+
run_optimization()
|
259 |
+
gr.Info("Optimization Finished! Move to the next step.")
|
260 |
+
return "Optimization finished! Move to the next step."#,gr.Button("Step 3. Run Editing",interactive = True)
|
261 |
+
except Exception as e:
|
262 |
+
print(e)
|
263 |
+
gr.Error("e")
|
264 |
+
return "Error: use a smaller batch size or try latter."#,gr.Button("Step 3. Run Editing",interactive = False)
|
265 |
+
|
266 |
+
|
267 |
+
|
268 |
+
if 1:
|
269 |
+
with gr.Row():
|
270 |
+
with gr.Column():
|
271 |
+
canvas_text_edit = gr.Image(value = None, type = "pil", label="Editing results", show_label=True,visible = True)
|
272 |
+
# canvas_text_edit = gr.Gallery(label = "Edited results")
|
273 |
+
|
274 |
+
with gr.Column():
|
275 |
+
gr.Markdown("""<p style="text-align: center; font-size: 20px">Editing setting</p>""")
|
276 |
+
tgt_prompt = gr.Textbox(value="text prompt", label="Editing: Text prompt", interactive= True )
|
277 |
with gr.Accordion(label="Advanced settings", open=False):
|
278 |
+
slider2 = gr.Slider(0, 20, step=1, label = 'mask id', visible=False)
|
279 |
+
guidance_scale = gr.Textbox(value="5", label="Editing: CFG guidance scale", interactive= True )
|
280 |
+
num_sampling_steps = gr.Number(value="20", label="Editing: Sampling steps", interactive= True )
|
281 |
+
edge_thickness = gr.Number(value="10", label="Editing: Edge thickness", interactive= True )
|
282 |
+
strength = gr.Textbox(value="0.5", label="Editing: Mask strength", interactive= True )
|
|
|
|
|
|
|
|
|
|
|
283 |
|
284 |
+
add_button = gr.Button("Step 2. Run Editing",interactive = True)
|
285 |
+
def run_edit_text_wrapper(
|
286 |
mask_np_list,
|
287 |
mask_label_list,
|
288 |
image,
|
|
|
289 |
num_tokens,
|
290 |
+
guidance_scale,
|
291 |
+
num_sampling_steps ,
|
292 |
+
strength ,
|
293 |
+
edge_thickness,
|
294 |
+
tgt_prompt ,
|
295 |
+
tgt_index
|
296 |
):
|
297 |
+
|
298 |
+
run_edit_text = partial(
|
299 |
+
run_main,
|
300 |
+
mask_np_list=mask_np_list,
|
301 |
+
mask_label_list=mask_label_list,
|
302 |
+
image_gt=np.array(image),
|
303 |
+
load_trained=True,
|
304 |
+
text=True,
|
305 |
+
num_tokens = int(num_tokens_global.value),
|
306 |
+
guidance_scale = float(guidance_scale),
|
307 |
+
num_sampling_steps = int(num_sampling_steps),
|
308 |
+
strength = float(strength),
|
309 |
+
edge_thickness = int(edge_thickness),
|
310 |
+
num_imgs = 1,
|
311 |
+
tgt_prompt = tgt_prompt,
|
312 |
+
tgt_index = int(tgt_index)
|
313 |
+
)
|
314 |
+
run_edit_text()
|
315 |
+
gr.Info('Image editing completed.')
|
316 |
+
return load_pil_img()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
317 |
|
318 |
|
319 |
|