SSTVOS: Sparse Spatiotemporal Transformers for Video Object Segmentation
- Resource Type
- Conference
- Authors
- Duke, Brendan; Ahmed, Abdalla; Wolf, Christian; Aarabi, Parham; Taylor, Graham W.
- Source
- 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) CVPR Computer Vision and Pattern Recognition (CVPR), 2021 IEEE/CVF Conference on. :5908-5917 Jun, 2021
- Subject
- Computing and Processing
Computer vision
Codes
Runtime
Scalability
Computational modeling
Object segmentation
Transformers
- Language
- ISSN
- 2575-7075
In this paper we introduce a Transformer-based approach to video object segmentation (VOS). To address compounding error and scalability issues of prior work, we propose a scalable, end-to-end method for VOS called Sparse Spatiotemporal Transformers (SST). SST extracts per-pixel representations for each object in a video using sparse attention over spatiotemporal features. Our attention-based formulation for VOS allows a model to learn to attend over a history of multiple frames and provides suitable inductive bias for performing correspondence-like computations necessary for solving motion segmentation. We demonstrate the effectiveness of attention-based over recurrent networks in the spatiotemporal domain. Our method achieves competitive results on YouTube-VOS and DAVIS 2017 with improved scalability and robustness to occlusions compared with the state of the art. Code is available at https://github.com/dukebw/SSTVOS.