Update app.py

app.py

@@ -29,70 +29,59 @@ def frame_capture(video_path, num_frames=5):
     vidObj.release()
     return frames
 
-# Function for railway defect analysis with a set prompt
-def railway_defect_analysis(video_path):
+# Function to generate text descriptions for frames
+def generate_descriptions_for_frames(video_path):
     frames = frame_capture(video_path)
     images = [PIL.Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)) for frame in frames]
     
-    prompt = "Describe any potential railway defects or issues visible in these video frames."
+    prompt = "Describe what is happening in each of these frames in this video sequentially."
     images_with_prompt = [prompt] + images
     
     responses = model.generate_content(images_with_prompt)
     descriptions = [response.text for response in responses]
     
-    return format_descriptions(descriptions)
+    formatted_description = format_descriptions(descriptions)
+    return formatted_description
 
-# Function to generate text descriptions for frames or answer a specific question
-def analyze_video(video_path, user_question):
+# Function to handle chat interaction
+def chat_interaction(video_path, chatbot, user_message):
     frames = frame_capture(video_path)
     images = [PIL.Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)) for frame in frames]
     
-    if user_question.strip():
-        prompt = f"Based on these video frames, {user_question}"
-    else:
-        prompt = "Describe what is happening in each of these frames in this video sequentially."
-    
+    prompt = f"Based on these video frames, {user_message}"
     images_with_prompt = [prompt] + images
     
     responses = model.generate_content(images_with_prompt)
-    descriptions = [response.text for response in responses]
+    response_text = responses[-1].text if user_message.strip() else format_descriptions([r.text for r in responses])
     
-    return descriptions[-1] if user_question.strip() else format_descriptions(descriptions)
+    chatbot.append((user_message, response_text))
+    return "", chatbot
 
 # Helper function to format descriptions
 def format_descriptions(descriptions):
     return ' '.join(descriptions).strip()
 
-# Function to handle chat interaction
-def chat_interaction(video_path, chatbot, user_message):
-    response = analyze_video(video_path, user_message)
-    chatbot.append((user_message, response))
-    return "", chatbot
-
-# Define Gradio interface
-# Separate video inputs for each tab to avoid the DuplicateBlockError
-railway_video_input = gr.Video(label="Upload Video for Railway Analysis", autoplay=True)
-chat_video_input = gr.Video(label="Upload Video for Chat Analysis", autoplay=True)
-
-# Railway defect analysis tab
-with gr.Blocks() as railway_tab:
+# Define the Gradio interfaces for each tab
+# Tab 1: Video Analysis System with Set Prompt
+with gr.Blocks() as tab1:
     with gr.Column():
-        gr.Markdown("### Railway Defect Analysis")
-        defect_analysis_output = gr.Textbox(label="Defect Analysis")
-        analyze_button = gr.Button("Analyze for Railway Defects")
-        analyze_button.click(fn=railway_defect_analysis, inputs=railway_video_input, outputs=defect_analysis_output)
+        gr.Markdown("### Video Analysis System")
+        video_input_1 = gr.Video(label="Upload Video", autoplay=True)
+        output_text = gr.Textbox(label="What's in this video")
+        analyze_button = gr.Button("Analyze Video")
+        analyze_button.click(fn=generate_descriptions_for_frames, inputs=video_input_1, outputs=output_text)
 
-# Interactive chat tab
-with gr.Blocks() as chat_tab:
+# Tab 2: Interactive Chat Mode
+with gr.Blocks() as tab2:
     with gr.Column():
         gr.Markdown("### Interactive Chat Mode")
+        video_input_2 = gr.Video(label="Upload Video", autoplay=True)
         chatbot = gr.Chatbot(label="Video Analysis Chatbot")
         user_input = gr.Textbox(label="Ask something specific about the video", placeholder="E.g., Are there any cars in this video?")
-        user_input.submit(fn=chat_interaction, inputs=[chat_video_input, chatbot, user_input], outputs=[user_input, chatbot])
+        user_input.submit(fn=chat_interaction, inputs=[video_input_2, chatbot, user_input], outputs=[user_input, chatbot])
 
 # Combine the two tabs into a single interface
 with gr.Blocks() as demo:
-    with gr.TabbedInterface([railway_tab, chat_tab], ["Railway Defect Analysis", "Interactive Chat"]):
-        pass
+    gr.TabbedInterface([tab1, tab2], ["Video Analysis", "Interactive Chat"])
 
 demo.launch()