跳到主要导航 跳到搜索 跳到主要内容

IoU-Enhanced Attention for End-to-End Task Specific Object Detection

  • Shanghai Key Laboratory of Multidimensional Information Processing

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

摘要

Without densely tiled anchor boxes or grid points in the image, sparse R-CNN achieves promising results through a set of object queries and proposal boxes updated in the cascaded training manner. However, due to the sparse nature and the one-to-one relation between the query and its attending region, it heavily depends on the self attention, which is usually inaccurate in the early training stage. Moreover, in a scene of dense objects, the object query interacts with many irrelevant ones, reducing its uniqueness and harming the performance. This paper proposes to use IoU between different boxes as a prior for the value routing in self attention. The original attention matrix multiplies the same size matrix computed from the IoU of proposal boxes, and they determine the routing scheme so that the irrelevant features can be suppressed. Furthermore, to accurately extract features for both classification and regression, we add two lightweight projection heads to provide the dynamic channel masks based on object query, and they multiply with the output from dynamic convs, making the results suitable for the two different tasks. We validate the proposed scheme on different datasets, including MS-COCO and CrowdHuman, showing that it significantly improves the performance and increases the model convergence speed. Codes are available at https://github.com/bravezzzzzz/IoU-Enhanced-Attention.

源语言英语
主期刊名Computer Vision – ACCV 2022 - 16th Asian Conference on Computer Vision, Proceedings
编辑Lei Wang, Juergen Gall, Tat-Jun Chin, Imari Sato, Rama Chellappa
出版商Springer Science and Business Media Deutschland GmbH
页124-141
页数18
ISBN(印刷版)9783031263477
DOI
出版状态已出版 - 2023
活动16th Asian Conference on Computer Vision, ACCV 2022 - Hybrid, Macao, 中国
期限: 4 12月 2022 → 8 12月 2022

丛书

姓名Lecture Notes in Computer Science
卷13845 LNCS
ISSN(印刷版)0302-9743
ISSN(电子版)1611-3349

会议

会议16th Asian Conference on Computer Vision, ACCV 2022
国家/地区中国
市Hybrid, Macao
时期4/12/22 → 8/12/22

学术指纹

探究 'IoU-Enhanced Attention for End-to-End Task Specific Object Detection' 的科研主题。它们共同构成独一无二的学术指纹。

引用此