用supervision在YOLOv11 上实现目标跟踪

import numpy as np
import supervision as sv
from ultralytics import YOLO

model = YOLO(r"E:\leilanfang\ultralytics-yolo11-main\yolo11n.onnx")
tracker = sv.ByteTrack()
box_annotator = sv.BoxAnnotator()
label_annotator = sv.LabelAnnotator()

def callback(frame: np.ndarray, _: int) -> np.ndarray:
    results = model(frame)[0]
    detections = sv.Detections.from_ultralytics(results)
    detections = tracker.update_with_detections(detections)

    labels = [
        f"#{tracker_id} {class_name}"
        for class_name, tracker_id
        in zip(detections.data["class_name"], detections.tracker_id)
    ]

    annotated_frame = box_annotator.annotate(
        frame.copy(), detections=detections)
    return label_annotator.annotate(
        annotated_frame, detections=detections, labels=labels)

sv.process_video(
    source_path=r"E:\leilanfang\ultralytics-yolo11-main\3.mp4",
    target_path="result.mp4",
    callback=callback
)