VALSE 论文速览 第88期：MSG-Transformer: 基于信使Token的高效视觉Transformer ...

论文题目：MSG-Transformer: Exchanging Local Spatial Information by Manipulating Messenger Tokens

作者列表：Jiemin Fang (HUST), Lingxi Xie (Huawei), Xinggang Wang (HUST), Xiaopeng Zhang (Huawei), Wenyu Liu (HUST), Qi Tian (Huawei)

B站观看网址：

https://www.bilibili.com/video/BV1vT41177RK/

论文摘要：

Transformers have offered a new methodology of designing neural networks for visual recognition. Compared to convolutional networks, Transformers enjoy the ability of referring to global features at each stage, yet the attention module brings higher computational overhead that obstructs the application of Transformers to process high-resolution visual data. This paper aims to alleviate the conflict between efficiency and flexibility, for which we propose a specialized token for each region that serves as a messenger (MSG). Hence, by manipulating these MSG tokens, one can flexibly exchange visual information across regions and the computational complexity is reduced. We then integrate the MSG token into a multi-scale architecture named MSG-Transformer. In standard image classification and object detection, MSG-Transformer achieves competitive performance and the inference on both GPU and CPU is accelerated. Code is available at https://github.com/hustvl/MSG-Transformer.

论文信息：

[1] Jiemin Fang, Lingxi Xie, Xinggang Wang, Xiaopeng Zhang, Wenyu Liu, Qi Tian: Proceedings of the IEEE/ CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, pp. 12063-12072.

论文链接：

[https://arxiv.org/abs/2105.15168]

代码链接：

[https://github.com/hustvl/MSG-Transformer]

视频讲者简介：

方杰民，华中科技大学电子信息与通信学院博士生，主要研究方向集中在基于神经网络架构设计的高效深度学习和神经渲染。在相关领域的国际期刊/ 会议 (IEEE TPAMI, CVPR, ICLR, ICCV, NeurIPS等)上发表多篇论文。

个人主页：

https://jaminfong.cn

特别鸣谢本次论文速览主要组织者：

月度轮值AC：王智慧 (大连理工大学)、杨旭 (西安电子科技大学)

季度责任AC：魏秀参 (南京理工大学)