MscaVPR: Multi-Scale Coordinate Attention Network for Robust Visual Place Recognition

Xiaohan Gao, Zhinong Zhong, Yongjian Tan, Ning Jing, Anran Yang, Qingren Jia · 2026

Read MscaVPR: Multi-Scale Coordinate Attention Network for Robust Visual Place Recognition by Xiaohan Gao, Zhinong Zhong, Yongjian Tan, Ning Jing, Anran Yang, Qingren Jia (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13211274.txt.

Description

A robot recognizing a place it has seen before must survive changed light and angles, and a 2026 computer vision study led by Xiaohan Gao strengthens that skill. The research article presents MscaVPR, a network for visual place recognition that fuses features across scales and learns from the direction each photo faces.

Pyramid attention gathers regional detail at several resolutions, while coordinate attention keeps positional cues alive through the layers. Training deliberately picks hard matches photographed from very different viewpoints and adds an azimuth-aware penalty to guide learning from them. Across benchmark sets the method improved top-one recall over a strong baseline, with the clearest gains on difficult night and sun datasets.

Catalog record

CREATORXiaohan Gao, Zhinong Zhong, Yongjian Tan, Ning Jing, Anran Yang, Qingren Jia
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE40 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC13211274

Browse the full library · About the Tower of Babel