Can I use cuda7.5 rather than cuda8.0 about deformable-convnets HOT 7 CLOSED

I think it is OK. Just compile a MXNet w/ cuda 7.5.

from deformable-convnets.

@orpine Finally I installed the environment successfully(Cuda7.5, Cudnn5.1.3, Python2.7.13 by anaconda). But I found another problem that when I run python ./rfcn/demo.py , I had the following error:

Segmentation fault (core dumped)

Could you have me?

from deformable-convnets.

@orpine how can i do step 3 from Installation:

Copy operators in ./rfcn/operator_cxx to $(YOUR_MXNET_FOLDER)/src/operator/contrib and recompile MXNet.

if I use mxnet in virtualenv?

from deformable-convnets.

@orpine I solved the problem by #3 (comment)

from deformable-convnets.

@zhFuECL How did you solve the problem? Could you give a detail?

from deformable-convnets.

@baiyancheng20 I gave the solution in the last comment.

from deformable-convnets.

@zhFuECL could you give a detail since there are many suggestions in #3 (comment)

from deformable-convnets.