Update app.py

app.py

@@ -159,11 +159,8 @@ def convert_video_to_mp4(input_path, output_path):
 def detect_objects_in_video(video_path):
     temp_output_path = "/tmp/output_video.mp4"
     temp_frames_dir = tempfile.mkdtemp()
-    
-    all_class_count = {}  # To store cumulative counts for all classes (from previous frames)
-    last_frame_objects = {}  # To store the objects detected in the last frame
     frame_count = 0
-    count_text = ""  # Initialize count_text properly
+    previous_detections = {}  # Untuk menyimpan deteksi objek dari frame sebelumnya
 
     try:
         # Convert video to MP4 if necessary
@@ -195,35 +192,46 @@ def detect_objects_in_video(video_path):
             # Process predictions for the current frame
             predictions = yolo_model.predict(frame_path, confidence=50, overlap=80).json()
 
-            # Get current frame object counts
-            current_frame_objects = {}
+            # Track current frame detections
+            current_detections = {}
             for prediction in predictions['predictions']:
                 class_name = prediction['class']
-                current_frame_objects[class_name] = current_frame_objects.get(class_name, 0) + 1
-
-                # Draw bounding boxes on the frame
                 x, y, w, h = prediction['x'], prediction['y'], prediction['width'], prediction['height']
-                cv2.rectangle(frame, (int(x-w/2), int(y-h/2)), (int(x+w/2), int(y+h/2)), (0,255,0), 2)
-                cv2.putText(frame, class_name, (int(x-w/2), int(y-h/2-10)), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,255,0), 2)
+                # Generate a unique ID for each detection (can use coordinates or other method)
+                object_id = f"{class_name}_{x}_{y}"
 
-            # Update all class counts: Add objects detected in this frame and remove objects not detected in this frame
-            for class_name, count in current_frame_objects.items():
-                # Add to all_class_count
-                all_class_count[class_name] = all_class_count.get(class_name, 0) + count
+                if object_id not in current_detections:
+                    current_detections[object_id] = class_name
 
-            # Now check for objects that were in last frame but not in the current frame
-            for class_name in list(all_class_count.keys()):
-                # If the object was not detected in the current frame, reduce its count
-                if class_name not in current_frame_objects:
-                    all_class_count[class_name] -= 1
-                    if all_class_count[class_name] <= 0:
-                        del all_class_count[class_name]  # Remove from all_class_count if count drops to 0
+                # Draw bounding box for detected objects
+                cv2.rectangle(frame, (int(x-w/2), int(y-h/2)), (int(x+w/2), int(y+h/2)), (0,255,0), 2)
+                cv2.putText(frame, class_name, (int(x-w/2), int(y-h/2-10)), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,255,0), 2)
 
-            # Update count_text with current object count
+            # Calculate the changes from previous detections
+            removed_objects = set(previous_detections.keys()) - set(current_detections.keys())
+            new_objects = set(current_detections.keys()) - set(previous_detections.keys())
+
+            # Update counts for objects
+            object_counts = {}
+            for detection_id in current_detections.keys():
+                class_name = current_detections[detection_id]
+                object_counts[class_name] = object_counts.get(class_name, 0) + 1
+
+            # Update object counts based on removed objects
+            for detection_id in removed_objects:
+                class_name = previous_detections[detection_id]
+                if class_name in object_counts:
+                    object_counts[class_name] -= 1
+                    if object_counts[class_name] <= 0:
+                        del object_counts[class_name]
+
+            # Generate display text for counts
             count_text = ""
-            for class_name, count in all_class_count.items():
+            total_product_count = 0
+            for class_name, count in object_counts.items():
                 count_text += f"{class_name}: {count}\n"
-            count_text += f"\nTotal Products: {sum(all_class_count.values())}"
+                total_product_count += count
+            count_text += f"\nTotal Product: {total_product_count}"
 
             # Overlay the counts text onto the frame
             y_offset = 20
@@ -235,6 +243,9 @@ def detect_objects_in_video(video_path):
             output_video.write(frame)
             frame_count += 1
 
+            # Update previous_detections for the next frame
+            previous_detections = current_detections
+
         video.release()
         output_video.release()