/dvtb-cpu

A docker container for deep visualization toolbox that runs with CPU only

Primary LanguagePython

This is a docker container for https://github.com/yosinski/deep-visualization-toolbox to get the toolbox running with CPU only without going through the hassle of manual installation.

Please check out https://github.com/fishcorn/dvtb-container for a docker container of the same toolbox but runs with CUDA GPUs. I created this container because fishcorn/dvtb-container did not work on my MacBook which lacks Nvidia GPU.

Running this container

First, pull the image from the docker-hub using:

docker pull mhalsharif/dvtb-cpu

To enable X server connection between the host and the container, run:

xhost +

On Linux (video cam input is supported), run this container with:

docker run -ti --name dvtb-container --device=/dev/video0:/dev/video0 --env="DISPLAY" --net=host -v /tmp/.X11-unix:/tmp/.X11-unix mhalsharif/dvtb-cpu

On macOS (video cam input is not supported), run this container with:

docker run -ti --name dvtb-container -e DISPLAY=host.docker.internal:0 -v /tmp/.X11-unix:/tmp/.X11-unix mhalsharif/dvtb-cpu

Before running the toolbox, make sure that you have downloaded (fetched) one of the models in the 'models' folder. For instance:

cd models
cd caffenet-yos
./fetch.sh

Finally, run the toolbox in /opt/dvtb folder with:

./run_toolbox.py

Building the image

First clone this repository:

git clone https://github.com/mhalsharif/dvtb-cpu.git

Then build the image with:

docker build -t dvtb-cpu .