code base of object detection.
1. 安装conda python环境
- `conda create -n xxx python=3.7/3.8`
- `conda activate xxx`
2. 运行脚本,自动安装pytorch1.8.0环境
- `bash install.sh'
1. 修改`configs`里的模型配置文件,包括数据集路径,类别等;
2. 运行`train.py`训练,设置config,epochs,gpu等:
- `python train.py -c configs/xxx.yaml --epochs * --n_gpu *`
`python val.py -c configs/xxx.yaml --weights (训练完成的模型权重)`
`python detect.py -c configs/xxx.yaml --weights (权重) -s (voc/cam)`
- YOLOv5: https://github.com/ultralytics/yolov5
- MMDetection:https://github.com/open-mmlab/mmdetection
- DBB: https://github.com/DingXiaoH/DiverseBranchBlock
- RepVGG:https://github.com/DingXiaoH/RepVGG
- RetinaNet:https://github.com/zhenghao977/RetinaNet-Pytorch-36.4AP