fix

app.py

@@ -21,8 +21,14 @@ def fn(video: gr.Video):
   print('Process video.')
   output = f'Marked-{str(video)}'
   clip = VideoFileClip(video)
-  data_dir = 'storage/dataset'
-  id_to_name, _ = ClassMapping(data_dir)
+  id_to_name = {
+    0: 'Flying',
+    1: 'Landing',
+    2: 'Other',
+    3: 'Straight Taxiing',
+    4: 'Takeoff',
+    5: 'Turning Maneuver',
+  }
   process_frame = FrameProcessor(detector, classifier, id_to_name)
   clip = clip.fl_image(process_frame)
   clip.write_videofile(output, fps=clip.fps, audio_codec='aac', logger=None)

core/inference.py

@@ -15,10 +15,6 @@ def detect_object(detector, frame):
   result = detector(frame, classes=4, verbose=False)[0]
   classes = result.boxes.cls.numpy()
   boxes = result.boxes.xyxy.numpy()
-  predictions = [
-    (result.names[classes[i]].capitalize(), boxes[i])
-    for i in range(len(classes))
-  ]
   detections = (
     [result.names[i].capitalize() for i in classes],
     boxes,
@@ -35,7 +31,7 @@ def classify_action(classifier, frames, id_to_name):
   for (class_id, confidence) in enumerate(confidences):
     other_class_id = 2
     if confidence > 0.3 and class_id != other_class_id:
-      actions.append(f'{id_to_name[class_id]}: {np.round(confidence, 2)}')
+      actions.append(f'{id_to_name[class_id]}: {confidence:.2f}')
   return actions
 
 def draw_boxes(frame, detections, actions):