search

rohitgirdhar / ActionVLAD

ActionVLAD for video action classification (CVPR 2017)
https://rohitgirdhar.github.io/ActionVLAD/
Other
216 stars 61 forks source link

Adding docker with python2, cuda8 #27

Closed fninaparavecino closed 5 years ago

fninaparavecino commented 5 years ago

Adding a docker that will solve the dependencies issues. It is working with --runtime=nvidia docker to access GPU. You will need cudnn and weights of the network in a docker_files folder

rohitgirdhar commented 5 years ago

Thanks @fninaparavecino! Could you please add a couple lines to the README with some instructions on how to use it?

fninaparavecino commented 5 years ago

@rohitgirdhar you can build the docker with $ docker build -t action:latest .

There is the need to have docker_files folder where there should be the cudnn5.1 (include and lib) and also the models folder.

ajay-Bhammar commented 5 years ago

@rohitgirdhar you can build the docker with $ docker build -t action:latest .

There is the need to have docker_files folder where there should be the cudnn5.1 (include and lib) and also the models folder.

I have built docker successfully. Can you tell next instruction(like how to run docker image etc.) ? because when I run python file it showing some python libraries missing which is docker image. I am new to docker. Am I doing something wrong?

fninaparavecino commented 5 years ago

@ajay-Bhammar to run docker image use the next line: $ docker run -it --runtime=nvidia action:latest The Dockerfile is running with nvidia docker so it can use the GPU. Once within the docker you have to run the regular way for the demo of ActionVLAD the code should be in the /app folder

ajay-Bhammar commented 5 years ago

@ajay-Bhammar to run docker image use the next line: $ docker run -it --runtime=nvidia action:latest The Dockerfile is running with nvidia docker so it can use the GPU. Once within the docker you have to run the regular way for the demo of ActionVLAD the code should be in the /app folder

Thanks for help