Update app.py

app.py

@@ -1,6 +1,6 @@
 import gradio as gr
 import pixeltable as pxt
-from pixeltable.functions.huggingface import clip
+from pixeltable.functions.huggingface import clip_image, clip_text
 from pixeltable.iterators import FrameIterator
 import PIL.Image
 import os
@@ -13,7 +13,7 @@ def process_video(video_file, progress=gr.Progress()):
     pxt.drop_dir('video_search', force=True)
     pxt.create_dir('video_search')
 
-    # Update type declaration to use simpler syntax
+    # Create the video table
     video_table = pxt.create_table('video_search.videos', {'video': pxt.Video})
 
     frames_view = pxt.create_view(
@@ -26,11 +26,11 @@ def process_video(video_file, progress=gr.Progress()):
     video_table.insert([{'video': video_file.name}])
     
     progress(0.4, desc="Creating embedding index...")
-    # Fixed parameter names for the embedding index
+    # Use separate methods for image and text embedding
     frames_view.add_embedding_index(
         'frame',
-        image_embed=clip.using(model_id='openai/clip-vit-base-patch32'),
-        string_embed=clip.using(model_id='openai/clip-vit-base-patch32')
+        image_embed=clip_image.using(model_id='openai/clip-vit-base-patch32'),
+        string_embed=clip_text.using(model_id='openai/clip-vit-base-patch32')
     )
 
     progress(1.0, desc="Processing complete")
@@ -41,52 +41,24 @@ def similarity_search(query, search_type, num_results, progress=gr.Progress()):
     frames_view = pxt.get_table('video_search.frames')
     
     progress(0.5, desc="Performing search...")
-    if search_type == "Text":
-        sim = frames_view.frame.similarity(query)
-    else:  # Image search
-        sim = frames_view.frame.similarity(query)
+    # The query is the same either way, but this makes it clearer what's happening
+    sim = frames_view.frame.similarity(query)
     
     results = frames_view.order_by(sim, asc=False).limit(num_results).select(frames_view.frame, sim=sim).collect()
     
     progress(1.0, desc="Search complete")
     return [row['frame'] for row in results]
     
-# Gradio interface
+# Simplified gradio interface with minimal styling
 with gr.Blocks() as demo:
-    gr.Markdown(
-        """
-        <div style="margin-bottom: 20px;">
-            <img src="https://raw.githubusercontent.com/pixeltable/pixeltable/main/docs/resources/pixeltable-logo-large.png" alt="Pixeltable" style="max-width: 150px;" />
-            <h2>Text and Image similarity search on video frames with embedding indexes</h2>
-        </div>
-        """
-    )
-    gr.HTML(
-    """
-    <p>
-        <a href="https://github.com/pixeltable/pixeltable" target="_blank" style="color: #F25022; text-decoration: none; font-weight: bold;">Pixeltable</a> is a declarative interface for working with text, images, embeddings, and even video, enabling you to store, transform, index, and iterate on data.
-    </p>
-    """
-    )
-
+    gr.Markdown("# Text and Image Search on Video Frames with Pixeltable")
     
     with gr.Row():
         with gr.Column(scale=1):
-
-            gr.Markdown(
-            """
-            <h3>1. Insert video</h3>
-            """)
-            
             video_file = gr.File(label="Upload Video")
             process_button = gr.Button("Process Video")
             process_output = gr.Textbox(label="Status", lines=2)
             
-            gr.Markdown(
-            """
-            <h3>2. Search video frames</h3>
-            """)
-            
             search_type = gr.Radio(["Text", "Image"], label="Search Type", value="Text")
             text_input = gr.Textbox(label="Text Query")
             image_input = gr.Image(label="Image Query", type="pil", visible=False)
@@ -94,24 +66,7 @@ with gr.Blocks() as demo:
             search_button = gr.Button("Search")
         
         with gr.Column(scale=2):
-
-            gr.Markdown(
-            """
-            <h3>3. Visualize results</h3>
-            """)
-            
             results_gallery = gr.Gallery(label="Search Results", columns=3)
-       
-            gr.Examples(
-            examples=[
-                ["bangkok.mp4"],
-                ["lotr.mp4"],
-                ["mi.mp4"],
-            ],
-            label="Click one of the examples below to get started",
-            inputs=[video_file],
-            fn=process_video
-            )
     
     def update_search_input(choice):
         return gr.update(visible=choice=="Text"), gr.update(visible=choice=="Image")
@@ -134,6 +89,6 @@ with gr.Blocks() as demo:
         outputs=[results_gallery]
     )
 
+# Launch with share=True to fix the localhost issue
 if __name__ == "__main__":
-    # Removed theme parameter which might be causing issues
-    demo.launch()
+    demo.launch(share=True)