二 Yolo源码训练AI模型

二 Yolo源码训练AI模型1 配置环境安装Python安装AI编程IDE环境软件-Trae CNpycharm安装教程python之安装使用Jupyter NoteBookPytorch和Cuda安装安装Yolo查看Yolo版本importultralyticsprint(ultralytics.__version__)yolo26发布于v8.4.0更新到最新Yolo版本pip install-U ultralytics--user2. 验证Yolo是否可用fromultralytics import YOLO# Load a modelmodelYOLO(rModels\yolo11s.pt)# load an official model# Predict with the modelresultsmodel(cat1.jpg,saveTrue)# predictonan image# Access the resultsforresultinresults:xywhresult.boxes.xywh # center-x,center-y,width,heightxywhnresult.boxes.xywhn #normalizedxyxyresult.boxes.xyxy # top-left-x,top-left-y,bottom-right-x,bottom-right-yxyxynresult.boxes.xyxyn #normalizednames[result.names[cls.item()]forclsinresult.boxes.cls.int()]#classnameof eachboxconfsresult.boxes.conf # confidence score of each box3.训练检测模型fromultralytics import YOLO# 0 参数配置# 模型路径model_pathr./Models/yolo26s.pt# 数据集yaml文件路径data_yamlrdet_data.yaml# 训练轮数epochs30imgsz224batch4projectr./AIModelnamePG_Detpredict_ImgPathrD:\dataset\PGDataset\images\valsave_predictImg_flagTrueexportTypeonnxexist_okTrue# 1 加载模型modelYOLO(model_path)# 2 训练模型model.train(datadata_yaml,epochsepochs,imgszimgsz,batchbatch,workers0,projectproject,namename,exist_okTrue)# 3 验证模型metricsmodel.val(datadata_yaml)# 在验证集上评估模型性能# 4 模型预测resultsmodel.predict(sourcepredict_ImgPath,imgszimgsz,savesave_predictImg_flag,batchbatch)# save plotted images# 5 导出所需模式以onnx为例model.export(formatexportType,imgszimgsz)