win10下亲测有效如果想在tensorrtcuda下部署yolov8直接看第五5章yolov8 官方仓库: https://github.com/ultralytics/ultralytics目录一、win10下创建yolov8环境二、推理图像、视频、摄像头2.1 推理图片2.2 推理视频2.3 推理摄像头三、训练3.1 快速训练coco128数据集3.2 预测四、导出onnx五、yolov8的tensorrt部署加速附录一、win10下创建yolov8环境# 注python其他版本在win10下可能有坑我已经替你踩坑了这里python3.9亲测有效 conda create -n yolov8 python3.9 -y conda activate yolov8 pip install ultralytics -i https://pypi.tuna.tsinghua.edu.cn/simple二、推理图像、视频、摄像头模型下载地址# download offical weights(.pt file) https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8n.pt https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8s.pt https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8m.pt https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8l.pt https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8x.pt https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8x6.pt这里下载yolov8n使用该模型推理图片原图如下图我们将图像和yolov8n.pt放到路径d:/Data/2.1 推理图片yolo modepredict modeld:/Data/yolov8n.pt sourced:/Data/6406407.jpg save show效果图默认保存在路径ultralytics\runs\detect\predict效果如下2.2 推理视频直接把source改成视频文件路径即可命令如下yolo modepredict modeld:/Data/yolov8n.pt sourced:/Data/people.mp4 save showbash窗口打印如下2.3 推理摄像头指令如下yolo modepredict modeld:/Data/yolov8n.pt source0 show三、训练3.1 快速训练coco128数据集在win10下创建路径D:\CodePython\yolov8将这个5Mb的数据集下载并解压在目录coco128数据集下载地址(别担心免费白嫖)文件分享如下图新建train.py文件代码如下from ultralytics import YOLO # Load a model # yaml会自动下载 model YOLO(yolov8n.yaml) # build a new model from scratch model YOLO(d:/Data/yolov8n.pt) # load a pretrained model (recommended for training) # Train the model results model.train(datacoco128.yaml, epochs100, imgsz640)训练指令python train.py如下图训练状态3.2 预测新建predict.py文件代码如下from ultralytics import YOLO # Load a model model YOLO(d:/Data/yolov8n.pt) # load an official model # Predict with the model results model(d:/Data/6406407.jpg, saveTrue) # predict on an image预测指令python predict.py如下图预测窗口打印图片默认保存在 D:\CodePython\ultralytics\runs\detect\predict四、导出onnxpip install onnx yolo modeexport modeld:/Data/yolov8n.pt formatonnx dynamicTrue opset12五、yolov8的tensorrt部署加速《YOLOV8部署保姆教程》 https://blog.csdn.net/m0_72734364/article/details/128758544?spm1001.2014.3001.5501​​​​​​TensorRT-Alpha基于tensorrtcuda c实现模型end2end的gpu加速支持win10、linux在2023年已经更新模型YOLOv8, YOLOv7, YOLOv6, YOLOv5, YOLOv4, YOLOv3, YOLOX, YOLOR,pphumanseg,u2net,EfficientDet。TensorRT-AlphaGitHub - FeiYull/TensorRT-Alpha at legacy · GitHubTensorRT系列之 Windows10下yolov8 tensorrt模型加速部署https://blog.csdn.net/m0_72734364/article/details/128865904TensorRT系列教程之 Linux下 yolov8 tensorrt模型加速部署https://blog.csdn.net/m0_72734364/article/details/128758544TensorRT系列教程之 Linux下 yolov7 tensorrt模型加速部署https://blog.csdn.net/m0_72734364/article/details/128759430TensorRT系列教程之 Linux下 yolov6 tensorrt模型加速部署https://blog.csdn.net/m0_72734364/article/details/133771238TensorRT系列教程之 Linux下 yolov5 tensorrt模型加速部署https://blog.csdn.net/m0_72734364/article/details/133773842TensorRT系列教程之 Linux下 yolox tensorrt模型加速部署https://blog.csdn.net/m0_72734364/article/details/133851320TensorRT系列教程之 Linux下 u2net tensorrt模型加速部署https://blog.csdn.net/m0_72734364/article/details/133887663?spm1001.2014.3001.5501快速看看yolov8n 在移动端RTX2070m(8G)的新能表现modelvideo resolutionmodel input sizeGPU Memory-UsageGPU-Utilyolov8n1920x10808x3x640x6401093MiB/7982MiB14%下图是yolov8n的运行时间开销单位是ms更多TensorRT-Alpha测试录像在B站视频B站YOLOv8nB站YOLOv8s附录更多训练指引请看官方文档。# yolov8 官方仓库: https://github.com/ultralytics/ultralytics # yolov8 官方中文教程ultralytics/README.zh-CN.md at main · ultralytics/ultralytics · GitHub # yolov8 官方训练指引: https://docs.ultralytics.com/reference/base_trainer/ # yolov8 官方快速教程: https://docs.ultralytics.com/quickstart/