Learning Instance Occlusion

This is the code for the CVPR 2020 paper "Learning Instance Occlusion for Panoptic Segmentation".

This project is based off of the excellent maskrcnn-benchmark. We extend it to:

1. Support Panoptic Segmentation/Panoptic FPN
2. Support learning an instance-wise relationship to determine occlusion
3. Integrate this into the existing greedy merging heuristic

If you make use of the ideas or code in this project, please consider citing:

@InProceedings{Lazarow_2020_CVPR,
author = {Lazarow, Justin and Lee, Kwonjoon and Shi, Kunyu and Tu, Zhuowen},
title = {Learning Instance Occlusion for Panoptic Segmentation},
booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2020}
}