Towards building comprehensive real-world visual perception systems, we propose and study a new problem called panoptic scene graph generation (PVSG). PVSG relates to the existing video scene graph generation (VidSGG) problem, which focuses on temporal interactions between humans and objects grounded with bounding boxes in videos. However, the limitation of bounding boxes in detecting non-rigid objects and backgrounds often causes VidSGG to miss key details crucial for comprehensive video understanding. In contrast, PVSG requires nodes in scene graphs to be grounded by more precise, pixel-level segmentation masks, which facilitate holistic scene understanding. To advance research in this new area, we contribute the PVSG dataset, which consists of 400 videos (289 third-person + 111 egocentric videos) with a total of 150K frames labeled with panoptic segmentation masks as well as fine, temporal scene graphs. We also provide a variety of baseline methods and share useful design practices for future work.
@article{arxiv.2311.17058,
title = {Panoptic Video Scene Graph Generation},
author = {Jingkang Yang and Wenxuan Peng and Xiangtai Li and Zujin Guo and Liangyu Chen and Bo Li and Zheng Ma and Kaiyang Zhou and Wayne Zhang and Chen Change Loy and Ziwei Liu},
journal= {arXiv preprint arXiv:2311.17058},
year = {2023}
}
Comments
Accepted to CVPR 2023. Project Page: https://jingkang50.github.io/PVSG/. Codebase: https://github.com/LilyDaytoy/OpenPVSG. We provide 400 long videos with frame-level panoptic segmentation, scene graph, dense captions, and QA annotations