path-tree
root|
- cocoapi
- maskrcnn
进入虚拟环境
pipenv shell
环境依赖
Cython==0.29.7
ipython==7.5.0
Keras==2.0.8
matplotlib==3.1.0
opencv-python==4.1.0.25
scikit-image==0.15.0
tensorflow-gpu==1.13.1
scipy==1.2.1
step 1 编译setup.py
window版本: extra_compile_args=[] # originally was ['-Wno-cpp', '-Wno-unused-function', '-std=c99'],
ubuntu版本: extra_compile_args=['-Wno-cpp', '-Wno-unused-function', '-std=c99'],
- pycocotools 库
cd ./cocoapi/PythonAPI
pip install Cython
python setup.py build_ext --inplace
python setup.py build_ext install
- copy pycocotools to maskrcnn 复制./cocoapi/PythonAPI/pycocotools 到 maskrcnn 文件夹中
- install pipy_whl 工具包
matplotlib
scipy==1.2.1
keras==2.0.8
scikit-image{skimage}
IPython
opencv-python
step 2 权重h5
mask_rcnn_coco_humanpose.h5
运行检测
cd maskrcnn
python video_demo.py