Deformable-ConvNets
Deformable-ConvNets copied to clipboard
Problem in "python ./rfcn/demo.py" operator _zeros is not implemented for GPU
I have already complished all procedure, but some problem occurs in running "python ./rfcn/demo.py"
ERROR: operator _zeros is not implemented for GPU But, I have already "make -j $(nproc) USE_OPENCV=1 USE_BLAS=openblas USE_CUDA=1 USE_CUDA_PATH=/usr/local/cuda USE_CUDNN=1"
thanks for help.
Seems your mxnet is compiled without cuda?
@Oh233 but i have already "make -j $(nproc) USE_OPENCV=1 USE_BLAS=openblas USE_CUDA=1 USE_CUDA_PATH=/usr/local/cuda USE_CUDNN=1" , USE_CUDA=1 . is there any question?
You can install mxnet by pip install ...
Have you solved this problem? If you have solved this problem, can you share the solution with me?
Have you solved this problem? If you have solved this problem, can you share the solution with me?
I met this problem too when installing mxnet as FGFA. And I had set USE_CUDA=1. This is because there are many versions of mxnet in my server and I guess python uses the wrong version. So what you need to do is :
- check if you have many mxnet versions: In my server, the path of mxnet is "/large2/zhaowei/anaconda3/envs/manet3/lib/python2.7/site-packages/mxnet-0.10.0-py2.7.egg". You should check if there are other versions of mxnet in "/large2/zhaowei/anaconda3/envs/manet3/lib/python2.7/site-packages".
- remove the other versions using "rm -r"
- then python demo.py should work