Calgary-Campinas challenge

This folder contains the training code specific for the Calgary Campinas challenge. As of writing (25 Oct 2020) this is the top result in both Track 1 and Track 2.

Training

The standard training script train_rim.py in tools/ can be used. If you want, the validation volumes can be computed using predict_val.py. During training, validation metrics will be logged, these match the challenge metrics. For our submission we used base.yaml as model configuration.

After downloading the data to <data_root> a command such as the one below was used (running in the docker container, which maps the code to direct):

cd /direct/tools
python train_rim.py <data_root>/Train/ \
                    <data_root>/Val/ \
                    <output_folder> \
                    --name base \
                    --cfg /direct/projects/calgary_campinas/configs/base.yaml \
                    --num-gpus 4 \
                    --num-workers 8 \
                    --resume

Additional options can be found using python train_rim.py --help.

Prediction

The masks are not provided for the test set, and need to be pre-computed using compute_masks.py. These masks should be passed to the --masks parameter of predict_test.py.