trackers目标跟踪库安装和试用

给算法爸爸上香2026-04-22 8:17

环境搭建

bash 复制代码

git clone https://github.com/roboflow/trackers.git
conda create -n trackers python=3.12.0
conda activate trackers
pip install torch==2.7.0 --index-url https://download.pytorch.org/whl/cu128               
pip install inference -i https://pypi.mirrors.ustc.edu.cn/simple/   
pip install -e . -i https://pypi.mirrors.ustc.edu.cn/simple/

demo测试

测试demo：

python 复制代码

import cv2
import supervision as sv
from inference import get_model
from trackers import ByteTrackTracker

model = get_model(model_id="rfdetr-medium")
tracker = ByteTrackTracker()

label_annotator = sv.LabelAnnotator()
trajectory_annotator = sv.TraceAnnotator()

cap = cv2.VideoCapture("720p60hz.mp4")

while cap.isOpened():
    ret, frame = cap.read()
    if not ret:
        break

    result = model.infer(frame)[0]
    detections = sv.Detections.from_inference(result)
    tracked = tracker.update(detections)

    frame = label_annotator.annotate(frame, tracked)
    frame = trajectory_annotator.annotate(frame, tracked)
    
    cv2.imshow("frame", frame)
    cv2.waitKey(1)

效果如下：

如果检测速度很慢，说明模型在cpu端运行。若需要在gpu上运行模型，需要执行：

bash 复制代码

pip uninstall onnxruntime
pip install onnxruntime-gpu -i https://pypi.mirrors.ustc.edu.cn/simple/