Explicit Time-Frequency Dynamics for Skeleton-Based Gait Recognition

Seoyeon Ko, Yeojin Song, Egene Chung, Luca Quagliato, Taeyong Lee, Junhyug Noh · 2026

Read Explicit Time-Frequency Dynamics for Skeleton-Based Gait Recognition by Seoyeon Ko, Yeojin Song, Egene Chung, Luca Quagliato, Taeyong Lee, Junhyug Noh (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-2604-03002.txt.

Description

Ko and five colleagues add explicit motion dynamics to skeleton-based gait recognition. Their Wavelet Feature Stream turns joint velocity sequences into multi-scale scalograms via wavelet transform. A lightweight multi-scale CNN then learns dynamic cues from them. The descriptor fuses with the backbone representation, with no backbone changes or extra supervision. On CASIA-B, the stream lifts strong backbones and sets a new skeleton-based best with GaitMixer. Gains run largest when subjects carry bags or wear coats.

Catalog record

CREATORSeoyeon Ko, Yeojin Song, Egene Chung, Luca Quagliato, Taeyong Lee, Junhyug Noh
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE17308
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.03002

Browse the full library · About the Tower of Babel