cdeep3m-docker

Docker container for running CDEEP3m - CUDA 9.0 - OpenCV 2 - CuDNN 7

Dependencies:

1. docker & docker-compose
2. nvidia-docker

To run execute the following steps:

1. Collect and prepare training data

mkdir train
cp -r 'TRAININGFILES AND FOLDERS' train/

2. Build docker compose

docker-compose build 

3. Edit and replace COMMAND in docker-compose.yml

version: '2.3'
services:
  cdeep3m:
    build:
      dockerfile: Dockerfile
      context: .
    image: cdeep3m
    runtime: nvidia
    environment:
      - NVIDIA_VISIBLE_DEVICES=all
    command: COMMAND # runtraining.sh COMMAND 
    volumes:
      - ./train:/train

4. Run:

docker-compose up

5. If rebuilding docker-compose image after changes to cdeep3m master run:

docker-compose build --no-cache