一. cpu版demo
1. 下载 tf-faster-rcnn
git clone https://github.com/endernewton/tf-faster-rcnn.git
2.为了匹配本机的GPU,更新setup中的-arch,参照以下表格.(只验证了cpu版的)
cd tf-faster-rcnn/lib
# Change the GPU architecture (-arch) if necessary
vim setup.py
| GPU model | Architecture |
|---|---|
| TitanX (Maxwell/Pascal) | sm_52 |
| GTX 960M | sm_50 |
| GTX 1080 (Ti) | sm_61 |
| Grid K520 (AWS g2.2xlarge) | sm_30 |
| Tesla K80 (AWS p2.xlarge) | sm_37 |
3.编译Cpython
make clean
make
cd ..
注意:
报错 ImportError: ../Lib/nms/gpu_nms.so: undefined symbol: _Py_ZeroStruct
原因:Makefile文件中 python setup.py 与 python demo.py版本不对应
4.准备数据
情况一:
/tf-faster-rcnn-master/data/demo
