This repository contains a docker image for rop classification (No rop, pre-plus and plus) from vessel segmented images.
$ docker pull docker.pkg.github.com/qtim-lab/rop_classification/rop:v0
FROM docker.pkg.github.com/qtim-lab/rop_classification/rop:v0
Loading into the container will automatically set the working directory to src/ which contains all the code.
However, the dataset (segmented vessels) must be mounted while running the container.
Below is an example:
nvidia-docker run --rm -it -v {Absolute path to segmented images in local system}:/segmented -v {Path to code}:/src rop_classification:v0
Once inside the container, to load a pre-trained model and run inference, run:
cd src
python3 main.py
Note: The above script will load a pre-trained model and will allow training for 1 epoch. Further parameters can be changed by modifying main.py
Authors:
- Ashwin Vaswani
- Katharina Hoebel
- Praveer Singh