期刊
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021
卷 -, 期 -, 页码 14035-14044出版社
IEEE COMPUTER SOC
DOI: 10.1109/CVPR46437.2021.01382
关键词
-
资金
- National Natural Science Foundation of China [61972162]
- CCF-Tencent Open Research fund
This paper introduces a novel Domain Agent Network (DAN) for Few-Shot Video Object Segmentation (FSVOS) task, breaking down full-rank attention into two smaller ones to improve performance. By considering one single frame of the query video as the domain agent, DAN effectively bridges between support images and query videos, while a learning strategy combining meta-learning with online learning further enhances segmentation accuracy, achieving state-of-the-art performance on both computational cost and accuracy.
This paper tackles the task of Few-Shot Video Object Segmentation (FSVOS), i.e., segmenting objects in the query videos with certain class specified in a few labeled support images. The key is to model the relationship between the query videos and the support images for propagating the object information. This is a many-to-many problem and often relies on full-rank attention, which is computationally intensive. In this paper, we propose a novel Domain Agent Network (DAN), breaking down the full-rank attention into two smaller ones. We consider one single frame of the query video as the domain agent, bridging between the support images and the query video. Our DAN allows a linear space and time complexity as opposed to the original quadratic form with no loss of performance. In addition, we introduce a learning strategy by combining meta-learning with online learning to further improve the segmentation accuracy. We build a FSVOS benchmark on the Youtube-VIS dataset and conduct experiments to demonstrate that our method outperforms baselines on both computational cost and accuracy, achieving the state-of-the-art performance. Code is available at https://github.com/scutpaul/DANet.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据