YouMVOS: An Actor-Centric Multi-Shot Video Object Segmentation Dataset

YouMVOS: An Actor-Centric Multi-Shot Video Object Segmentation Dataset

Wei D, Kharbanda S, Arora S, Roy R, Jain N, Palrecha A, Shah T, Mathur S, Mathur R, Kemkar A, Chakravarthy A, Lin Z, Jang W, Tang Y, Bai S, Tompkin J, Torr PH, and Pfister H.

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022.

Many video understanding tasks require analyzing multi-shot videos, but existing datasets for video object segmentation (VOS) only consider single-shot videos. To address this challenge, we collected a new dataset---YouMVOS---of 200 popular YouTube videos spanning ten genres, where each video is on average five minutes long and with 75 shots. We selected recurring actors and annotated 431K segmentation masks at a frame rate of six, exceeding previous datasets in average video duration, object variation, and narrative structure complexity. We incorporated good practices of model architecture design, memory management, and multi-shot tracking into an existing video segmentation method to build competitive baseline methods. Through error analysis, we found that these baselines still fail to cope with cross-shot appearance variation on our YouMVOS dataset. Thus, our dataset poses new challenges in multi-shot segmentation towards better video analysis. Data, code, and pre-trained models are available.


This work has been supported by NSF grants NCS-FO2124179, NIH grant R01HD104969, UKRI grant Turing AI Fellowship EP/W002981/1, andE PSRC/MURI grant EP/N019474/1. We also thank the Royal Academy of Engineering and Five AI.