Swin-Transformer-Object-Detection(mmdetection) 安裝

 conda create -n stod -y python=3.8

conda activate stod

conda install pytorch=1.6.0 torchvision cudatoolkit=10.2 -c pytorch -y (可以替換自己需要的版本)

 

# mmcv安装

cd ~/software/

rm -rf mmcv

git clone https://github.com/open-mmlab/mmcv.git

cd mmcv

MMCV_WITH_OPS=1 pip install -e .

or

pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/{cu_version}/{torch_version}/index.html

eg: pip install mmcv-full=1.4.6 -f https://download.openmmlab.com/mmcv/dist/cu113/torch1.10.0/index.html

實際版本參考網站：https://mmcv.readthedocs.io/en/latest/get_started/installation.html

 

# 安装apex

git clone https://github.com/NVIDIA/apex

cd apex

pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./

 可能會安裝失敗

可以試試：python setup.py install

# mmdetection安装

git clone https://github.com/open-mmlab/mmdetection.git

cd mmdetection

pip install -r requirements/build.txt

pip install -r requirements.txt

pip install -v -e . 

or

pip install mmdet==(version)

注意：mmcv與mmdet的版本必須要搭配，版本參考：https://mmdetection.readthedocs.io/en/latest/get_started.html

 

cd /home/zengxh/workspace/Swin-Transformer-Object-Detection

pip install -r requirements.txt

or 

pip install -v -e .

————————————————

測試環境：

python demo/image_demo.py demo/demo.jpg configs/swin/mask_rcnn_swin_tiny_patch4_window7_mstrain_480-800_adamw_3x_coco.py mask_rcnn_swin_tiny_patch4_window7.pth

測試成功的環境：

mmcv-full: 1.3.18

mmdet: 2.11.0

pytorch: 1.6.0

cuda:10.2

or

gpu: 3060ti

mmcv-full: 1.4.6

mmdet: 2.22.0

pytorch: 1.8

cuda: 11.3

中間遇到提：

ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

原因：numpy版本太舊

pip install --upgrade numpy 解決