Introduction¶

This repository contains code and tools for reading, processing, evaluating on, and visualizing Panoptic Parts datasets. Moreover, it contains code for reproducing our CVPR 2021 paper results.

Datasets¶

Cityscapes-Panoptic-Parts and PASCAL-Panoptic-Parts are created by extending two established datasets for image scene understanding, namely Cityscapes and PASCAL datasets. Detailed description of the datasets and various statistics are presented in our technical report in arxiv. The datasets can be downloaded from:

Cityscapes Panoptic Parts
PASCAL Panoptic Parts (alternative link, code: i7ap)

API and code reference¶

We provide a public, stable API, and various code utilities that are documented here.

Reproducing CVPR 2021 paper¶

The part-aware panoptic segmentation results from the paper can be reproduced using this guide.

Evaluation metrics¶

We provide two metrics for evaluating performance on Panoptic Parts datasets.

Part-aware Panoptic Quality (PartPQ): here.
Intersection over Union (IoU): TBA

Citations¶

Please cite us if you find our work useful or you use it in your research:

@inproceedings{degeus2021panopticparts,
    title = {Part-aware Panoptic Segmentation},
    author = {Daan de Geus and Panagiotis Meletis and Chenyang Lu and Xiaoxiao Wen and Gijs Dubbelman},
    booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    year = {2021}
}

@article{meletis2020panopticparts,
    title = {Cityscapes-Panoptic-Parts and PASCAL-Panoptic-Parts datasets for Scene Understanding},
    author = {Panagiotis Meletis and Xiaoxiao Wen and Chenyang Lu and Daan de Geus and Gijs Dubbelman},
    type = {Technical report},
    institution = {Eindhoven University of Technology},
    date = {16/04/2020},
    url = {https://github.com/tue-mps/panoptic_parts},
    eprint={2004.07944},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}