一、训练环境:
1.创建虚拟环境:conda create -n yolov8 python=3.8
2.激活虚拟环境:conda activate yolov8
3.安装pytorch:
教程上的:pip install torch==1.12.0+cu116 torchvision==0.13.0+cu116 --extra-index-url https://download.pytorch.org/whl/cu116 -i https://pypi.tuna.tsinghua.edu.cn/simple
Pytorch环境下的:
conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch-nightly -c nvidia
选择与Pytorch环境保持一致
4.安装ultralytics:pip install ultralytics
//5.安装依赖包:pip install -r requirements.txt -i https://pypi.mirrors.ustc.edu.cn/simple/
二、voc数据集环境:
1.从yolov5环境拷贝makeTxt.py、voc_label_urine_cell_231230.py文件放在
/home/kemove/gfh/AI-train/yolov8/ultralytics-main/ultralytics/ 目录下
2.图片放在/home/kemove/gfh/AI-train/yolov8/ultralytics-main/ultralytics/data/images/目录下
3.标签放在/home/kemove/gfh/AI-train/yolov8/ultralytics-main/ultralytics/data/Annotations/目录下
三、模型训练:
1.在/home/kemove/gfh/AI-train/yolov8/ultralytics-main/ultralytics/ 目录依次执行:
python3 makeTxt.py
python3 voc_label_urine_cell_231230.py
2.启动训练:
方式一:yolo train data=./cfg/datasets/VOC-urine_cell_20231230.yaml model=yolov8s.pt batch=-1
方式二:yolo train data=./cfg/datasets/VOC-urine_cell_20231230.yaml model=./cfg/models/v8/yolov8s.yaml batch=-1 amp=false
注意:
1../cfg/datasets/VOC-urine_cell_20231230.yaml中path最好用绝对路径,发现在85服务器和75服务器上的相对路径不一样;
2.voc_label_urine_cell_231230.py在74和75服务器上,写入图片地址时,要用“./”替换"data/",原因暂时未知。