Docker container wrapping a simple Caffe model trained on ImageNet. Tinker with a hosted demo!
First, install nvidia-docker. The image is available on Docker Hub as pavlov/nin-imagenet.
Once the image is running, you can navigate to the web UI on port 80.
$ git clone https://github.com/pavlovml/nin-imagenet.git
$ make build
$ make run
You can configure the port that the web UI is exposed on via the DOCKER_PORT environment variable.
nin-imagenet bundles mavenlin's pre-trained ImageNet model, which is released under a non-commercial license. The HTTP server is based on Caffe's example server, which is released under the BSD 2-Clause license. nin-imagenet itself is released under the BSD 3-Clause license.