Docker images for the Digital Mammography DREAM Challenge

Overview

This repository contains the Dockerfiles required to build the example images provided for the Digital Mammography DREAM Challenge.

Docker images

Frameworks

• [x] caffe-gpu (Repository, Website)
• [x] tensorflow-gpu (Repository, Website)
• [x] theano-gp (Repository, Website)
• [x] keras-gpu (Repository, Website)
• [x] matlab-runtime-gpu (Repository, Website)

Pre-processing

• [x] dm-preprocess-png: resizes and converts the images to PNG (Repository)
• [x] dm-preprocess-lmdb: resizes and saves the images to a Lightning Memory-mapped Database (Repository)
• [x] dm-preprocess-caffe: prepares the data before training a Caffe model (Repository)

Training

• [x] dm-train-caffe: trains an AlexNet or GoogLeNet model using Caffe (Repository)
• [ ] dm-train-tensorflow
• [ ] dm-train-keras

Set Environment Variables

TODO: Add description on how to set environment variables (e.g. RANDOM_SEED)