AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

A Self-Supervised Gait Encoding Approach With Locality-Awareness for 3D Skeleton Based Person Re-Identification

Haocong Rao, Siqi Wang, Xiping Hu, Mingkui Tan, Yi Guo, Jun Sheng Cheng, Xinwang Liu, Bin Hu

TPAMI 2021 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Contrastive versus non-contrastive theory · 105 citations (OpenAlex)

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 decoder (self-supervised reconstruction) to the LSTM encoder (ablation)GE · BIWIRank-1: 36.1 -> 41.5Improved
add reverse reconstruction as pretext task (ablation)GE+GD · BIWIRank-1: 41.5 -> 46.7Improved
add locality-aware attention (LAS) (ablation)GE+GD+Rev. Rec. · BIWIRank-1: 46.7 -> 57.7Improved
add locality-aware contrastive learning (CAGEs vs AGEs) (ablation)full model · BIWIRank-1: 59.1 -> 62.9Improved
use reverse reconstruction instead of prediction as pretext (ablation)self-supervised gait encoding · BIWIRank-1 (AP): 40.4 -> 62.9Improved
use reverse reconstruction instead of sorting as pretext (ablation)self-supervised gait encoding · BIWIRank-1 (AP): 55.7 -> 62.9Improved
combine all three pretext tasks (Rev. Rec. Plus) vs Rev. Rec. alone (ablation)CAGEs · KS20Rank-1 (AP): 86.9 -> 92.0Improved
combine all three pretext tasks vs Rev. Rec. alone (ablation)CAGEs · IAS-ARank-1 (AP): 60.1 -> 59.1Hurt
use average prediction (AP) instead of sequence-level concatenation (SC) (ablation)CAGEs Rev. Rec. · BIWIRank-1: 51.3 -> 62.9Improved
contrast non-adjacent sequences (interval 2) instead of adjacent (ablation)LCL · BIWIRank-1: 62.9 -> 61.1Hurt
raise LCL temperature from 0.1 to 1 (ablation)LCL · BIWIRank-1: 62.9 -> 60.6No clear effect
apply attention (BAS, MBAS, LAS) vs no attention (ablation)reverse reconstruction model · multiple datasetsRank-1: 1.0%-7.2% Rank-1Improved

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