Gal-Faster-RCNN

Faster RCNN for Radio Galaxy Detection

Installation

Instructions for installation can be found here.

Data preparation

Download and extract RadioGalaxyNET data from here. We expect the directory structure to be the following:

./RadioGalaxyNET/
  annotations/  # annotation json files
  train/    # train images
  val/      # val images
  test/     # test images

Training

To train on a single node with single gpu run:

python train.py

Evaluation

To evaluate on test images with a single GPU run:

python evaluate.py

License

MIT license.