CDeep3M Docker

Make sure you have Docker installed. The most recent docker images of CDeep3M Version 2 are available here: cdeep3m2-docker There are many different ways to run docker. Here are instructions to keep a docker container running.

docker images

REPOSITORY	TAG	IMAGE ID	CREATED	SIZE
ncmir/cdeep3m	latest	23c87acf9c04	19 hours ago	6.5GB

Start a docker container that keeps running:

docker run -td --network=host -v /data:/data --gpus all --entrypoint "/usr/bin/top" 23c87acf9c04

(-v mount volume /data from your workstation into /data in the docker container)

Show running docker containers:

docker ps

CONTAINER ID	IMAGE	COMMAND	CREATED	STATUS	PORTS	NAMES
a9eef89b480e	23c87acf9c04	"/usr/bin/top"	4 seconds ago			nervous_williams

Enter the running docker container:

docker exec -it a9eef89b480e /bin/bash

This is your start screen:

root@*******: /home/cdeep3m#

Ready to start!

To run a training (assuming you have matching images and labels in /imagefolder and /labelfolder):

PreprocessTrainingData.py /imagefolder /labelfolder /augmented runtraining.sh --numiterations 50000 /augmented /trainednet

Running a quick demo with demo data:

PreprocessTrainingData.py /home/cdeep3m/mito_testsample/training/images/ /home/cdeep3m/mito_testsample/training/labels/ /home/cdeep3m/mito_testsample/mito_test_augmented

runtraining.sh --numiterations 50 --snapshot_interval 10 /home/cdeep3m/mito_testsample/mito_test_augmented /home/cdeep3m/mito_testsample/demo_trained

To run a prediction:

runprediction.sh /trainednet /my_images /predictout

Further explanations regarding runprediction.sh

Once you are done processing. Leave the container. exit

To stop the container:

docker kill a9eef89b480e