win10 安裝 detectron2

全新的detectron2

  1. 基於pytorch重寫
  2. 加入了新的功能如:panoptic segmentation, densepose, Cascade R-CNN, rotated bounding boxes
  3. 爲輪子而生、爲開源而生
  4. 訓練更快

作爲一隻新入門的菜鳥一直奉行:復現,復現,復現,自我發揮=自找麻煩

  • 開始
  • 安裝VS2019(可以不卸載VS2017\VS2015)
  • 安裝CUDA10.1及與之配套的cudnn(可以不卸載CUDA10.0),安裝完成後可以使用一下命令檢驗:
nvidia-smi
  • 安裝Anaconda
  • 創建一個虛擬環境  
conda create -n detectron2py36 python=3.6.5
  • 進入虛擬環境
activate detectron2py36
  • 安裝pytorch,可前往pytorch官網,或者直接使用以下命令,建議使用國內鏡像
pip install torch===1.3.1 torchvision===0.4.2 -f https://download.pytorch.org/whl/torch_stable.html -i https://pypi.tuna.tsinghua.edu.cn/simple

 並作如下修改:

file1:

  {your evn path}\Lib\site-packages\torch\include\torch\csrc\jit\argument_spec.h

  example:

  {C:\Miniconda3\envs\py36}\Lib\site-packages\torch\include\torch\csrc\jit\argument_spec.h(190)

    static constexpr size_t DEPTH_LIMIT = 128;

      change to -->

    static const size_t DEPTH_LIMIT = 128;

file2:

  {your evn path}\Lib\site-packages\torch\include\pybind11\cast.h

  example:

  {C:\Miniconda3\envs\py36}\Lib\site-packages\torch\include\pybind11\cast.h(1449)

    explicit operator type&() { return *(this->value); }

      change to -->

    explicit operator type&() { return *((type*)this->value); }

 

  • 安裝fvcore:
pip install 'git+https://github.com/facebookresearch/fvcore'
git clone [email protected]:facebookresearch/detectron2.git

做如下修改:

for the "detectron2/layers/csrc/ROIAlign/ROIAlign_cuda.cu" and "detectron2/layers/csrc/ROIAlignRotated/ROIAlignRotated_cuda.cu",
change from each

dim3 grid(std::min(at::cuda::ATenCeilDiv(output_size, 512L), 4096L));
~
~
dim3 grid(std::min(at::cuda::ATenCeilDiv(grad.numel(), 512L), 4096L));
to

dim3 grid(std::min(at::cuda::ATenCeilDiv(static_cast<int64_t>(output_size), static_cast<int64_t>(512)), static_cast<int64_t>(4096)));
 ~
~
dim3 grid(std::min(at::cuda::ATenCeilDiv(static_cast<int64_t>(grad.numel()), static_cast<int64_t>(512)), static_cast<int64_t>(4096)));

 

編譯:

cd detectron2
python setup.py build develop
  • 完成,侵刪

Inference :

https://github.com/facebookresearch/detectron2

https://github.com/facebookresearch/detectron2/issues/9

https://github.com/philferriere/cocoapi

https://github.com/facebookresearch/detectron2/issues/9#issuecomment-546226799

發表評論
所有評論
還沒有人評論,想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章