fix frames as video

app.py

@@ -19,6 +19,10 @@ class ChaplinGradio:
         self.frame_interval = 1 / self.fps
         self.frame_compression = 25
         self.last_frame_time = time.time()
+        
+        # Frame buffer
+        self.frame_buffer = []
+        self.buffer_size = 16  # Number of frames to accumulate before processing
 
     def download_models(self):
         """Download required model files from HuggingFace"""
@@ -57,7 +61,7 @@ class ChaplinGradio:
         print("Model loaded successfully!")
 
     def process_frame(self, frame):
-        """Process a single frame with rate limiting and compression"""
+        """Process frames with buffering"""
         current_time = time.time()
         
         if current_time - self.last_frame_time < self.frame_interval:
@@ -69,50 +73,59 @@ class ChaplinGradio:
             return "No video input detected"
         
         try:
-            # Create temp directory if it doesn't exist
-            os.makedirs("temp", exist_ok=True)
-            
-            # Generate temporary video file path
-            temp_video = f"temp/frame_{time.time_ns()}.mp4"
-            
-            # Compress and save frame as video
-            frame_height, frame_width = frame.shape[:2]
-            out = cv2.VideoWriter(
-                temp_video,
-                cv2.VideoWriter_fourcc(*'mp4v'),
-                self.fps,
-                (frame_width, frame_height),
-                False  # isColor
-            )
-            
             # Convert frame to grayscale if it's not already
             if len(frame.shape) == 3:
                 frame = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)
-                
-            # Write frame to video
-            out.write(frame)
-            out.release()
             
-            # Process the video file using the pipeline
-            try:
-                predicted_text = self.vsr_model(temp_video)
+            # Add frame to buffer
+            self.frame_buffer.append(frame)
+            
+            # Only process when we have enough frames
+            if len(self.frame_buffer) >= self.buffer_size:
+                # Create temp directory if it doesn't exist
+                os.makedirs("temp", exist_ok=True)
+                
+                # Generate temporary video file path
+                temp_video = f"temp/frames_{time.time_ns()}.mp4"
+                
+                # Get frame dimensions from first frame
+                frame_height, frame_width = self.frame_buffer[0].shape[:2]
                 
-                # Clean up temp file
-                os.remove(temp_video)
+                # Create video writer
+                out = cv2.VideoWriter(
+                    temp_video,
+                    cv2.VideoWriter_fourcc(*'mp4v'),
+                    self.fps,
+                    (frame_width, frame_height),
+                    False  # isColor
+                )
                 
-                return predicted_text
+                # Write all frames to video
+                for f in self.frame_buffer:
+                    out.write(f)
+                out.release()
                 
-            except Exception as e:
-                print(f"Error during inference: {str(e)}")
-                return f"Error processing frame: {str(e)}"
+                # Clear buffer
+                self.frame_buffer = []
+                
+                try:
+                    # Process the video file using the pipeline
+                    predicted_text = self.vsr_model(temp_video)
+                    return predicted_text
+                    
+                except Exception as e:
+                    print(f"Error during inference: {str(e)}")
+                    return f"Error processing frames: {str(e)}"
+                finally:
+                    # Clean up temp file
+                    if os.path.exists(temp_video):
+                        os.remove(temp_video)
+            
+            return "Collecting frames..."  # Return status while collecting frames
                 
         except Exception as e:
-            print(f"Error saving frame: {str(e)}")
-            return f"Error saving frame: {str(e)}"
-        finally:
-            # Make sure we always try to clean up
-            if 'temp_video' in locals() and os.path.exists(temp_video):
-                os.remove(temp_video)
+            print(f"Error processing: {str(e)}")
+            return f"Error processing: {str(e)}"
 
 
 # Create Gradio interface