Visual Saliency Transformer

Nian Liu, Ni Zhang, Kaiyuan Wan, Ling Shao, Junwei Han

科研成果: 书/报告/会议事项章节会议稿件同行评审

361 引用 (Scopus)

摘要

Existing state-of-the-art saliency detection methods heavily rely on CNN-based architectures. Alternatively, we rethink this task from a convolution-free sequence-to-sequence perspective and predict saliency by modeling long-range dependencies, which can not be achieved by convolution. Specifically, we develop a novel unified model based on a pure transformer, namely, Visual Saliency Transformer (VST), for both RGB and RGB-D salient object detection (SOD). It takes image patches as inputs and leverages the transformer to propagate global contexts among image patches. Unlike conventional architectures used in Vision Transformer (ViT), we leverage multi-level token fusion and propose a new token upsampling method under the transformer framework to get high-resolution detection results. We also develop a token-based multi-task decoder to simultaneously perform saliency and boundary detection by introducing task-related tokens and a novel patch-task-attention mechanism. Experimental results show that our model outperforms existing methods on both RGB and RGB-D SOD benchmark datasets. Most importantly, our whole framework not only provides a new perspective for the SOD field but also shows a new paradigm for transformer-based dense prediction models. Code is available at https://github.com/nnizhang/VST.

源语言英语
主期刊名Proceedings - 2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021
出版商Institute of Electrical and Electronics Engineers Inc.
4702-4712
页数11
ISBN(电子版)9781665428125
DOI
出版状态已出版 - 2021
活动18th IEEE/CVF International Conference on Computer Vision, ICCV 2021 - Virtual, Online, 加拿大
期限: 11 10月 202117 10月 2021

出版系列

姓名Proceedings of the IEEE International Conference on Computer Vision
ISSN(印刷版)1550-5499

会议

会议18th IEEE/CVF International Conference on Computer Vision, ICCV 2021
国家/地区加拿大
Virtual, Online
时期11/10/2117/10/21

指纹

探究 'Visual Saliency Transformer' 的科研主题。它们共同构成独一无二的指纹。

引用此