Compilation of pre-trained deep learning models with demos and code.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.
containers Performance improvement. Nov 27, 2017
docs Deprecation warning. Aug 10, 2018
.gitignore Clean repo structure. Oct 23, 2017
.gitmodules Clean repo structure. Oct 23, 2017
LICENSE First commit. Sep 28, 2017 Deprecation warning. Aug 10, 2018


[DEPRECATED] Sortable and searchable compilation of pre-trained deep learning models. With demos and code.

DEPRECATED: You can find an alternative here

A word of warning

This is running on a server without GPU, hence it seems slow.

Also, the code may look a bit like monkey-patching for the following reasons:

  • Models are cloned as submodules: therefore we have to mess around with the python path :-(
  • There is a queuing systems for the jobs (allows user to see their job's position in the queue)


Having spent too much time installing deep learning models just to evaluate their performance, I created this repo for several reasons:

  • Access a free demo of deep learning models
  • Gather available deep learning models
  • Get a docker container running the model for a quick install


Requirements: Docker, docker-compose and enough space free for the model weights.

git clone --recursive
cd containers
docker-compose build
docker-compose up -d

Useful commands

docker ps #list images
docker attach [container_id] #attach a shell to specific image


Many models are missing. Any help is welcome ! You have two options to contribute.

Easy way: Add a model to the list without a demo:

  • Fork the repo
  • Edit the docs/models.yaml file (you can even edit it with Github's editor)
  • Make a pull request

Other way: add a model with a demo:

  • Fork the repo
  • Add the model inside one of the docker containers
  • Create a route in the file
  • Add a demo calling the route
  • Make a pull request


  • Use nvidia-docker ?
  • Add flag to compile Tensorflow
  • Consider splitting each model in a different container ?
  • Linter
  • Add analytics