当前位置: X-MOL 学术Inform. Fusion › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
PESA-Net: Permutation-Equivariant Split Attention Network for correspondence learning
Information Fusion ( IF 18.6 ) Pub Date : 2021-08-02 , DOI: 10.1016/j.inffus.2021.07.018
Zhen Zhong 1 , Guobao Xiao 1 , Shiping Wang 2 , Leyi Wei 3 , Xiaoqin Zhang 4
Affiliation  

Establishing reliable correspondences by a deep neural network is an important task in computer vision, and it generally requires permutation-equivariant architecture and rich contextual information. In this paper, we design a Permutation-Equivariant Split Attention Network (called PESA-Net), to gather rich contextual information for the feature matching task. Specifically, we propose a novel “Split–Squeeze–Excitation–Union” (SSEU) module. The SSEU module not only generates multiple paths to exploit the geometrical context of putative correspondences from different aspects, but also adaptively captures channel-wise global information by explicitly modeling the interdependencies between the channels of features. In addition, we further construct a block by fusing the SSEU module, Multi-Layer Perceptron and some normalizations. The proposed PESA-Net is able to effectively infer the probabilities of correspondences being inliers or outliers and simultaneously recover the relative pose by essential matrix. Experimental results demonstrate that the proposed PESA-Net relative surpasses state-of-the-art approaches for pose estimation and outlier rejection on both outdoor scenes and indoor scenes (i.e., YFCC100M and SUN3D). Source codes: https://github.com/x-gb/PESA-Net.



中文翻译:

PESA-Net:用于对应学习的置换等变分裂注意网络

通过深度神经网络建立可靠的对应关系是计算机视觉中的一项重要任务,它通常需要置换等变架构和丰富的上下文信息。在本文中,我们设计了一个 Permutation-Equivariant Split Attention Network(称为 PESA-Net),为特征匹配任务收集丰富的上下文信息。具体来说,我们提出了一种新颖的“Split-Squeeze-Excitation-Union”(SSEU)模块。SSEU 模块不仅生成多条路径以从不同方面利用假定对应的几何上下文,而且还通过显式建模特征通道之间的相互依赖性来自适应地捕获通道方面的全局信息。此外,我们通过融合 SSEU 模块、多层感知器和一些归一化进一步构建了一个块。提出的 PESA-Net 能够有效地推断出对应关系为内点或离群点的概率,并同时通过基本矩阵恢复相对位姿。实验结果表明,所提出的 PESA-Net 相对于室外场景和室内场景(即 YFCC100M 和 SUN3D)的姿态估计和异常值拒绝的最先进方法。源代码:https://github.com/x-gb/PESA-Net。

更新日期:2021-08-07
down
wechat
bug