基于TensorRT的C++高性能推理库，Yolov10, YoloPv2，Yolov5/7/X/8，RT-DETR，单目标跟踪OSTrack、LightTrack。

分别使用OpenCV、ONNXRuntime部署YOLOPV2目标检测+可驾驶区域分割+车道线分割，一共包含54个onnx模型，依然是包含C++和Python两个版本的程序。仅仅只依赖OpenCV就能运行，彻底摆脱对任何深度学习框架的依赖。

YOLOPv2のPythonでのONNX推論サンプル

Fine tuned YOLOPv2 model with borrowed features from the Utils library and depthai_python MobileNet human detection model to send both braking and steering commands for self driving using pre-trained weights.