mask_rcnn

Mask RCNN

How to train

Step 1: Set the correct PYTHONPATH

See the main README.md

Step 2: Run the training script

Use train.py to train a model. Configs we used are here. Download a pretrained model to speed up the training. We used the pretrained model for COCO. One thing to note is that this script splits the training data into train and validation sets in memory.

python train.py models/resnet50/config.ini \
    /path/to/train/dicoms \
    /path/to/train/label \
    /output/dir \
    --coco /path/to/coco/model.h5

Step 3: Evaluate and test

Use eval.py to evaluate or test the model.

# evaluate
python eval.py \
    models/resnet50/config.ini \
    0.1 \
    /path/to/dicom/images \
    /path/to/trained/model \
    /path/to/output/dir \
    --label-file /path/to/label/file \
    --pick-500

# test
# this is what we used for our submission
python eval.py \
    models/resnet50/config.ini \
    0.95 \
    /path/to/dicom/images \
    /path/to/trained/model \
    /path/to/output/dir \
    --submission-file /path/to/submission/file
python eval.py \
    models/resnet101/config.ini \
    0.98 \
    /path/to/dicom/images \
    /path/to/trained/model \
    /path/to/output/dir \
    --submission-file /path/to/submission/file