SwinTransformer 目标检测实践笔记

• SwinTransformer
• Swin-Transformer-Object-Detection
• bilibili
• beyonderwei

环境配置

conda create -n swin_det python=3.8

# To activate this environment, use
#     $ conda activate swin_det
# To deactivate an active environment, use
#     $ conda deactivate

pip3 install torch torchvision torchaudio

pip install cython matplotlib opencv-python timm -i http://mirrors.aliyun.com/pypi/simple/ --trusted-host mirrors.aliyun.com

1. mmcv

下载一个版本的zip,并移动到swin_det的目录下 (/opt/anaconda3/envs/swin_det)

cd /opt/anaconda3/envs/swin_det/mmcv-2.2.0

pip install -r requirements.txt

python setup.py build_ext

python setup.py develop

pip list 会看到 mmcv-full

2. mmdetection

下载一个版本的zip,并移动到swin_det的目录下 (/opt/anaconda3/envs/swin_det)

cd /opt/anaconda3/envs/swin_det/mmdetection-3.3.0

pip install -r requirements.txt -i http://mirrors.aliyun.com/pypi/simple/ --trusted-host mirrors.aliyun.com

python setup.py develop

Swin-Transformer-Object-Detection

下载源码并使用pycharm打开工程,设置python环境

conda create -n swin_det2 python=3.8

conda activate swin_det2

cd /Users/morningcat/Downloads/Swin-Transformer-Object-Detection-2.11.0

pip install -r requirements.txt

pip install torch==1.9.0

pip install mmcv-full==1.3.18

下载-权值文件：(直接丢在项目根目录下)

mask_rcnn_swin_tiny_patch4_window7.pth