NVIDIA / flownet2-pytorch

Pytorch implementation of FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks
Other
3.12k stars 739 forks source link

ModuleNotFoundError: No module named 'correlation_cuda? #95

Open long2double opened 5 years ago

long2double commented 5 years ago

Hello,I have a problem.looking forward to your reply。Thank you! After successfully executing the bash install.sh command, it shows why this error occurs during runtime. ModuleNotFoundError: No module named 'correlation_cuda? The compiler is pycharm.

BaranovMykola commented 5 years ago

Same problem

KittenMiao commented 5 years ago

try again with cuda 9.0

HPL123 commented 5 years ago

@changshuangzhao @BaranovMykola @kbrenneman ,Did you solve the problem?

BaranovMykola commented 5 years ago

@HPL123 No i didn't But i have avoided the problem by using docker for flownet

HPL123 commented 5 years ago

@BaranovMykola ,How to use docker?could you give me a guidance?thanks for your help.

BaranovMykola commented 5 years ago

@HPL123 Docker - it's similar to virtual machine, in a nutshell. You can pull docker image (prebuilded one) or create your own. Docker image is just a template for your docker container with installed stuff. Docker container is instance of your image. You can do everything what you want inside container and you never affect your primary OS.

My link above provides docker image for flownet2 with installed caffe, cuda, python examples

long2double commented 5 years ago

https://github.com/ClementPinard/FlowNetPytorch Hello, you can try this.

YukeLiang commented 4 years ago

Did you fix it? I have the same problem as well.