Update models/visual_analyzer.py

models/visual_analyzer.py

@@ -403,7 +403,7 @@ Respond with just the emotion."""
 
         Args:
             image: Image to analyze
-            prompt: Custom analysis prompt
+            prompt: Custom analysis prompt from user
             timestamp: Timestamp for this frame
 
         Returns:
@@ -419,7 +419,15 @@ Respond with just the emotion."""
         else:
             pil_image = image
 
-        response = self._query_model(pil_image, prompt, max_tokens=200)
+        # Format the prompt to get a clear yes/no style response for clip selection
+        formatted_prompt = (
+            f"Analyze this frame for the following criteria: {prompt}\n\n"
+            f"Does this frame match these criteria? "
+            f"Respond with 'Yes' or 'No' followed by a brief explanation. "
+            f"If it partially matches, say 'Partially' and explain what matches."
+        )
+
+        response = self._query_model(pil_image, formatted_prompt, max_tokens=200)
 
         return {
             "timestamp": timestamp,