注:制作数据集连接:制作自己的目标检测数据集-CSDN博客
训练数据集的步骤如下:
进入autodl中,开机后点击‘jupyterlab’,点击‘终端’
1.将数据集放在‘autodl-tmp’中,代码放在根目录下:
2.下载代码
git clone https://github.com/ultralytics/ultralytics
3..安装yolov8需要的依赖
cd ultralyitcs
pip install ultralytics
4.将下载好的预训练权重‘yolov8x.pt'放入’weights'
5.修改配置文件
#新建data.yaml文件
train: /root/autodl-tmp/person_data/images/train #训练集所在的路径
val: /root/autodl-tmp/person_data/images/val #训练时测试集所在的路径
test: /root/autodl-tmp/person_data/images/test
# number of classes
nc: 1 #类别
# class names
names: ['person']
#新建train.py
from ultralytics import YOLO
model = YOLO("yolov8x.pt")
results = model.train(data="data.yaml",imgsz=800, epochs=120, batch=16, device=0, workers=0)
6.训练
python train.py
训练完后,结果保存在‘ultralytics/runs’中
训练时出现的错误:
解决方式:train.py进行修改
from ultralytics import YOLO if __name__ == '__main__': model = YOLO('yolov8s.pt') results = model.train(data=r'/root/ultralytics/data.yaml', imgsz=800, epochs=120, batch=16, device='0', workers=0, )