Linoy Tsaban
commited on
Commit
•
d2ec8a3
1
Parent(s):
af2b22f
Update app.py
Browse files
app.py
CHANGED
@@ -31,7 +31,7 @@ def caption_image(input_image):
|
|
31 |
|
32 |
generated_ids = blip_model.generate(pixel_values=pixel_values, max_length=50)
|
33 |
generated_caption = blip_processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
34 |
-
return generated_caption
|
35 |
|
36 |
|
37 |
|
@@ -123,6 +123,7 @@ def load_and_invert(
|
|
123 |
def edit(input_image,
|
124 |
wts, zs,
|
125 |
tar_prompt,
|
|
|
126 |
steps,
|
127 |
skip,
|
128 |
tar_cfg_scale,
|
@@ -162,6 +163,8 @@ def edit(input_image,
|
|
162 |
eta=1,)
|
163 |
|
164 |
latnets = wts.value[skip].expand(1, -1, -1, -1)
|
|
|
|
|
165 |
sega_out = sem_pipe(prompt=tar_prompt, latents=latnets, guidance_scale = tar_cfg_scale,
|
166 |
num_images_per_prompt=1,
|
167 |
num_inference_steps=steps,
|
@@ -426,6 +429,9 @@ with gr.Blocks(css="style.css") as demo:
|
|
426 |
do_reconstruction = True
|
427 |
return do_reconstruction
|
428 |
|
|
|
|
|
|
|
429 |
def update_inversion_progress_visibility(input_image, do_inversion):
|
430 |
if do_inversion and not input_image is None:
|
431 |
return inversion_progress.update(visible=True)
|
@@ -446,6 +452,7 @@ with gr.Blocks(css="style.css") as demo:
|
|
446 |
do_inversion = gr.State(value=True)
|
447 |
do_reconstruction = gr.State(value=True)
|
448 |
sega_concepts_counter = gr.State(0)
|
|
|
449 |
|
450 |
|
451 |
|
@@ -659,6 +666,7 @@ with gr.Blocks(css="style.css") as demo:
|
|
659 |
inputs=[input_image,
|
660 |
wts, zs,
|
661 |
tar_prompt,
|
|
|
662 |
steps,
|
663 |
skip,
|
664 |
tar_cfg_scale,
|
@@ -689,7 +697,7 @@ with gr.Blocks(css="style.css") as demo:
|
|
689 |
outputs = [do_inversion],
|
690 |
queue = False).then(fn = caption_image,
|
691 |
inputs = [input_image],
|
692 |
-
outputs = [tar_prompt]).then(fn = update_inversion_progress_visibility, inputs =[input_image,do_inversion],
|
693 |
outputs=[inversion_progress],queue=False).then(
|
694 |
fn=load_and_invert,
|
695 |
inputs=[input_image,
|
|
|
31 |
|
32 |
generated_ids = blip_model.generate(pixel_values=pixel_values, max_length=50)
|
33 |
generated_caption = blip_processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
34 |
+
return generated_caption, generated_caption
|
35 |
|
36 |
|
37 |
|
|
|
123 |
def edit(input_image,
|
124 |
wts, zs,
|
125 |
tar_prompt,
|
126 |
+
image_caption,
|
127 |
steps,
|
128 |
skip,
|
129 |
tar_cfg_scale,
|
|
|
163 |
eta=1,)
|
164 |
|
165 |
latnets = wts.value[skip].expand(1, -1, -1, -1)
|
166 |
+
if image_caption == tar_prompt:
|
167 |
+
tar_prompt = ""
|
168 |
sega_out = sem_pipe(prompt=tar_prompt, latents=latnets, guidance_scale = tar_cfg_scale,
|
169 |
num_images_per_prompt=1,
|
170 |
num_inference_steps=steps,
|
|
|
429 |
do_reconstruction = True
|
430 |
return do_reconstruction
|
431 |
|
432 |
+
def reset_image_caption():
|
433 |
+
return ""
|
434 |
+
|
435 |
def update_inversion_progress_visibility(input_image, do_inversion):
|
436 |
if do_inversion and not input_image is None:
|
437 |
return inversion_progress.update(visible=True)
|
|
|
452 |
do_inversion = gr.State(value=True)
|
453 |
do_reconstruction = gr.State(value=True)
|
454 |
sega_concepts_counter = gr.State(0)
|
455 |
+
image_caption = gr.State(value="")
|
456 |
|
457 |
|
458 |
|
|
|
666 |
inputs=[input_image,
|
667 |
wts, zs,
|
668 |
tar_prompt,
|
669 |
+
image_caption,
|
670 |
steps,
|
671 |
skip,
|
672 |
tar_cfg_scale,
|
|
|
697 |
outputs = [do_inversion],
|
698 |
queue = False).then(fn = caption_image,
|
699 |
inputs = [input_image],
|
700 |
+
outputs = [tar_prompt, image_caption]).then(fn = update_inversion_progress_visibility, inputs =[input_image,do_inversion],
|
701 |
outputs=[inversion_progress],queue=False).then(
|
702 |
fn=load_and_invert,
|
703 |
inputs=[input_image,
|