Update app.py

app.py

@@ -1,17 +1,34 @@
 #!/usr/bin/env python3
 import gradio as gr
 from clip_interrogator import Config, Interrogator
+from share_btn import community_icon_html, loading_icon_html, share_js
 
-# MODELS = ['ViT-L (best for Stable Diffusion 1.*)', 'ViT-H (best for Stable Diffusion 2.*)']
-# MODELS = ['ViT-L (best for Stable Diffusion 1.*)',]
+MODELS = ['ViT-L (best for Stable Diffusion 1.*)']#, 'ViT-H (best for Stable Diffusion 2.*)']
 
 # load BLIP and ViT-L https://huggingface.co/openai/clip-vit-large-patch14
-from PIL import Image
-from clip_interrogator import Config, Interrogator
-
-ci = Interrogator(Config(clip_model_name="ViT-L-14/openai"))
+config = Config(clip_model_name="ViT-L-14/openai")
+ci_vitl = Interrogator(config)
+# ci_vitl.clip_model = ci_vitl.clip_model.to("cpu")
+
+# load ViT-H https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K
+# config.blip_model = ci_vitl.blip_model
+# config.clip_model_name = "ViT-H-14/laion2b_s32b_b79k"
+# ci_vith = Interrogator(config)
+# ci_vith.clip_model = ci_vith.clip_model.to("cpu")
+
+
+def image_analysis(image, clip_model_name):
+    # move selected model to GPU and other model to CPU
+    # if clip_model_name == MODELS[0]:
+    #     ci_vith.clip_model = ci_vith.clip_model.to("cpu")
+    #     ci_vitl.clip_model = ci_vitl.clip_model.to(ci_vitl.device)
+    #     ci = ci_vitl
+    # else:
+    #     ci_vitl.clip_model = ci_vitl.clip_model.to("cpu")
+    #     ci_vith.clip_model = ci_vith.clip_model.to(ci_vith.device)
+    #     ci = ci_vith
+    ci = ci_vitl
 
-def image_analysis(image):
     image = image.convert('RGB')
     image_features = ci.image_to_features(image)
 
@@ -30,7 +47,22 @@ def image_analysis(image):
     return medium_ranks, artist_ranks, movement_ranks, trending_ranks, flavor_ranks
 
 
-def image_to_prompt(image, mode):
+def image_to_prompt(image, clip_model_name, mode):
+    # move selected model to GPU and other model to CPU
+    # if clip_model_name == MODELS[0]:
+    #     ci_vith.clip_model = ci_vith.clip_model.to("cpu")
+    #     ci_vitl.clip_model = ci_vitl.clip_model.to(ci_vitl.device)
+    #     ci = ci_vitl
+    # else:
+    #     ci_vitl.clip_model = ci_vitl.clip_model.to("cpu")
+    #     ci_vith.clip_model = ci_vith.clip_model.to(ci_vith.device)
+    #     ci = ci_vith
+    ci = ci_vitl
+
+    ci.config.blip_num_beams = 64
+    ci.config.chunk_size = 2048
+    ci.config.flavor_intermediate_count = 2048 if clip_model_name == MODELS[0] else 1024
+
     image = image.convert('RGB')
     if mode == 'best':
         prompt = ci.interrogate(image)
@@ -41,7 +73,7 @@ def image_to_prompt(image, mode):
     elif mode == 'negative':
         prompt = ci.interrogate_negative(image)
 
-    return prompt
+    return prompt, gr.update(visible=True), gr.update(visible=True), gr.update(visible=True)
 
 
 TITLE = """
@@ -61,7 +93,7 @@ TITLE = """
         <p style="margin-bottom: 10px; font-size: 94%">
         Want to figure out what a good prompt might be to create new images like an existing one?<br>The CLIP Interrogator is here to get you answers!
         </p>
-        <p>You can skip the queue by duplicating this space and upgrading to gpu in settings: <a style='display:inline-block' href='https://huggingface.co/spaces/pharma/CLIP-Interrogator?duplicate=true'><img src='https://img.shields.io/badge/-Duplicate%20Space-blue?labelColor=white&style=flat&logo=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAP5JREFUOE+lk7FqAkEURY+ltunEgFXS2sZGIbXfEPdLlnxJyDdYB62sbbUKpLbVNhyYFzbrrA74YJlh9r079973psed0cvUD4A+4HoCjsA85X0Dfn/RBLBgBDxnQPfAEJgBY+A9gALA4tcbamSzS4xq4FOQAJgCDwV2CPKV8tZAJcAjMMkUe1vX+U+SMhfAJEHasQIWmXNN3abzDwHUrgcRGmYcgKe0bxrblHEB4E/pndMazNpSZGcsZdBlYJcEL9Afo75molJyM2FxmPgmgPqlWNLGfwZGG6UiyEvLzHYDmoPkDDiNm9JR9uboiONcBXrpY1qmgs21x1QwyZcpvxt9NS09PlsPAAAAAElFTkSuQmCC&logoWidth=14' alt='Duplicate Space'></a></p>
+        <p>You can skip the queue by duplicating this space and upgrading to gpu in settings: <a style='display:inline-block' href='https://huggingface.co/spaces/pharmapsychotic/CLIP-Interrogator?duplicate=true'><img src='https://img.shields.io/badge/-Duplicate%20Space-blue?labelColor=white&style=flat&logo=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAP5JREFUOE+lk7FqAkEURY+ltunEgFXS2sZGIbXfEPdLlnxJyDdYB62sbbUKpLbVNhyYFzbrrA74YJlh9r079973psed0cvUD4A+4HoCjsA85X0Dfn/RBLBgBDxnQPfAEJgBY+A9gALA4tcbamSzS4xq4FOQAJgCDwV2CPKV8tZAJcAjMMkUe1vX+U+SMhfAJEHasQIWmXNN3abzDwHUrgcRGmYcgKe0bxrblHEB4E/pndMazNpSZGcsZdBlYJcEL9Afo75molJyM2FxmPgmgPqlWNLGfwZGG6UiyEvLzHYDmoPkDDiNm9JR9uboiONcBXrpY1qmgs21x1QwyZcpvxt9NS09PlsPAAAAAElFTkSuQmCC&logoWidth=14' alt='Duplicate Space'></a></p>
     </div>
 """
 
@@ -72,11 +104,9 @@ ARTICLE = """
     and <a href="https://pixabay.com/illustrations/animal-painting-cat-feline-pet-7154059/">Lin Tong</a> 
     from pixabay.com
     </p>
-
     <p>
     Server busy? You can also run on <a href="https://colab.research.google.com/github/pharmapsychotic/clip-interrogator/blob/main/clip_interrogator.ipynb">Google Colab</a>
     </p>
-
     <p>
     Has this been helpful to you? Follow me on twitter 
     <a href="https://twitter.com/pharmapsychotic">@pharmapsychotic</a><br>
@@ -126,14 +156,14 @@ def analyze_tab():
             trending = gr.Label(label="Trending", num_top_classes=5)
             flavor = gr.Label(label="Flavor", num_top_classes=5)
 
-    button = gr.Button("Analyze")
+    button = gr.Button("Analyze", api_name="image-analysis")
     button.click(image_analysis, inputs=[image, model], outputs=[medium, artist, movement, trending, flavor])
 
     examples=[['example01.jpg', MODELS[0]], ['example02.jpg', MODELS[0]]]
     ex = gr.Examples(
         examples=examples, 
         fn=image_analysis, 
-        inputs=[input_image], 
+        inputs=[input_image, input_model], 
         outputs=[medium, artist, movement, trending, flavor], 
         cache_examples=True, 
         run_on_click=True
@@ -149,16 +179,22 @@ with gr.Blocks(css=CSS) as block:
             with gr.Row():
                 input_image = gr.Image(type='pil', elem_id="input-img")
                 with gr.Column():
+                    input_model = gr.Dropdown(MODELS, value=MODELS[0], label='CLIP Model')
                     input_mode = gr.Radio(['best', 'fast', 'classic', 'negative'], value='best', label='Mode')
-            submit_btn = gr.Button("Submit")
+            submit_btn = gr.Button("Submit", api_name="image-to-prompt")
             output_text = gr.Textbox(label="Output", elem_id="output-txt")
 
-            examples=[['example01.jpg', 'best'], ['example02.jpg', 'best']]
+            with gr.Group(elem_id="share-btn-container"):
+                community_icon = gr.HTML(community_icon_html, visible=False)
+                loading_icon = gr.HTML(loading_icon_html, visible=False)
+                share_button = gr.Button("Share to community", elem_id="share-btn", visible=False)
+
+            examples=[['example01.jpg', MODELS[0], 'best'], ['example02.jpg', MODELS[0], 'best']]
             ex = gr.Examples(
                 examples=examples, 
                 fn=image_to_prompt, 
-                inputs=[input_image, input_mode], 
-                outputs=[output_text], 
+                inputs=[input_image, input_model, input_mode], 
+                outputs=[output_text, share_button, community_icon, loading_icon], 
                 cache_examples=True, 
                 run_on_click=True
             )
@@ -171,8 +207,9 @@ with gr.Blocks(css=CSS) as block:
 
     submit_btn.click(
         fn=image_to_prompt, 
-        inputs=[input_image, input_mode], 
-        outputs=[output_text]
+        inputs=[input_image, input_model, input_mode], 
+        outputs=[output_text, share_button, community_icon, loading_icon]
     )
+    share_button.click(None, [], [], _js=share_js)
 
-block.queue(max_size=64).launch(show_api=False)
+block.queue(max_size=64).launch(show_api=False)