GaitKD: A Universal Decoupled Distillation Framework for Efficient Gait Recognition

Yuqi Li, Qian Zhou, Huiran Duan, Jingjie Wang. Shunli Zhang. Chuanguang Yang, Guoying Zhao, and Yingli Tian · 2026

Read GaitKD: A Universal Decoupled Distillation Framework for Efficient Gait Recognition by Yuqi Li, Qian Zhou, Huiran Duan, Jingjie Wang. Shunli Zhang. Chuanguang Yang, Guoying Zhao, and Yingli Tian (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/arxiv-2604-26255.txt.

Description

Li and seven colleagues built GaitKD, a teaching framework that helps small gait-recognition models learn from large ones. Walking style identifies people at a distance without contact, but strong models are often too heavy to run in practice. GaitKD splits the lesson in two: matching the teacher's decisions and keeping its boundary layout in the student's embedding space. The split design works across mixed teacher-student pairs and adds no cost at run time.

Catalog record

CREATORYuqi Li, Qian Zhou, Huiran Duan, Jingjie Wang. Shunli Zhang. Chuanguang Yang, Guoying Zhao, and Yingli Tian
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45321
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.26255

Browse the full library · About the Tower of Babel