AI Paper Atlas

Research map / Video / Action recognition

HyperGait: Unleashing the Power of Parsing for Gait Recognition in the Wild via Hypergraph

Jinkai Zheng, Jiaqing Wei, Xinxiang Jin, Yaoqi Sun, Xichun Sheng, Ming Li, Liangqiong Qu, Xinchen Liu, Wu Liu

CVPR 2026 · Video › Action recognition · approach: Skeleton-based action recognition

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add spatial hypergraph convolution module (SHCM) (ablation)Backbone + Global Head · Gait3DRank-1: 75.7 -> 76.7Improved
add temporal hypergraph convolution module (THCM) (ablation)Backbone + Global Head · Gait3DRank-1: 75.7 -> 77.9Improved
add both SHCM and THCM (ablation)Backbone + Global Head · Gait3DRank-1: 75.7 -> 80.5Improved
spatial hypergraph instead of spatial GCN (ablation)Baseline · Gait3DRank-1: 76.2 -> 76.7Improved
temporal hypergraph instead of temporal GCN (ablation)Baseline · Gait3DRank-1: 76.8 -> 77.9Improved
hypergraph modules instead of S-GCN + T-GCN combination (ablation)Baseline · Gait3DRank-1: 78.1 -> 80.5Improved
add spatial GCN to baseline (ablation)Backbone + Global Head · Gait3DmINP: 49.8 -> 47.3Hurt
spatial hypergraph threshold 0.4 vs 0.6 (ablation)HyperGait · Gait3DRank-1: 77.8 -> 80.5Improved
spatial hypergraph threshold 0.9 vs 0.4 (ablation)HyperGait · Gait3DRank-1: 80.5 -> 78.8Hurt
10 temporal segments instead of 5 (ablation)HyperGait · Gait3DRank-1: 79.3 -> 80.5Improved
increase temporal kNN from 3 to 7 (ablation)HyperGait · Gait3DRank-1: 80.5 -> 78.8Hurt
HyperGait on occlusion probe vs GaitHeatHyperGait · SUSTech1K occlusionRank-1: 53.2 vs 61.4Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy