/benchmark

Primary LanguagePython

NMS(in faster-rcnn) benchmark

1.ubuntu 18
2.python 3.7
3.cuda 9.2
4.gpu 2080ti

build _nms_gpu_post:

python build.py build_ext --inplace

build nms_cuda:

linux: nvcc --shared -Xcompiler -fPIC nms_cuda.cu nms_cuda_vision.cpp  -o nms_cuda.so  -I /home/fms/anaconda3/include/python3.7m    
windows: nvcc --shared -Xcompiler -fPIC nms_cuda.cu nms_cuda_vision.cpp  -o nms_cuda.pyd

or(pytorch 1.x):

python setup.py build_ext --inplace

run :

jupyter notebook

select:
benchmark.ipynb