Mask-R-CNN-with-Res2net-backbone

This repository contains the implementation of Mask R-CNN with Res2net backbone. The Res2net backbone code is part of the Res2net network which you can find in the paper titled "Res2Net: A New Multi-scale Backbone Architecture". This code was used with the RADIATE dataset available here: https://www.kaggle.com/competitions/data-science-bowl-2018/data.

Model

To run the code use the following command :

python3 train.py --epochs 100 --batch-size 2 --TRAIN_PATH "Data_Folder"