Trying to annotate a video, I’m using callback function from this blog.
predictions object looks like described in the post (https://fastupload.io/2a249668b939cb4f).
Here is my code:
job_id, signed_url, expire_time = model.predict_video(
"/content/IMG_7690.mp4",
fps=5,
prediction_type="batch-video",
additional_models = ["clip"]
)
results = model.poll_until_video_results(job_id)
def callback(scene: np.ndarray, index: int) -> np.ndarray:
result = results[index]
detections = sv.Detections.from_ultralytics(result)
bounding_box_annotator = sv.BoundingBoxAnnotator()
label_annotator = sv.LabelAnnotator()
labels = [
result.names[class_id]
for class_id
in detections.class_id
]
annotated_image = bounding_box_annotator.annotate(
scene=scene, detections=detections)
annotated_image = label_annotator.annotate(
scene=annotated_image, detections=detections, labels=labels)
return annotated_image
sv.process_video(
source_path="/content/IMG_7690.mp4",
target_path="output.mp4",
callback=callback
)
results are created before callback function, but I get the following error:
UnboundLocalError: local variable 'results' referenced before assignment
Renaming first ‘results’ variable to ‘test’ within callback gives me KeyError: 0.
def callback(scene: np.ndarray, index: int) -> np.ndarray:
test = results[index]
detections = sv.Detections.from_ultralytics(test)
bounding_box_annotator = sv.BoundingBoxAnnotator()
label_annotator = sv.LabelAnnotator()
labels = [
test.names[class_id]
for class_id
in detections.class_id
]
annotated_image = bounding_box_annotator.annotate(
scene=scene, detections=detections)
annotated_image = label_annotator.annotate(
scene=annotated_image, detections=detections, labels=labels)
return annotated_image
What can I do to complete video annotation?