AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Shape-Erased Feature Learning for Visible-Infrared Person Re-Identification

Jiawei Feng, Ancong Wu, Wei-Shi Zheng

CVPR 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Cross-modality and occlusion re-identification · 159 citations (OpenAlex)

Read the paper

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
Use shape-related objective without shape-erased objective (ablation)SGIEL · SYSU-MM01 all search, single-shotRank-1: 70.73 -> 69.60Hurt
Add shape-erased objective alongside shape-related objective (ablation)SGIEL · SYSU-MM01 all search, single-shotRank-1: 69.60 -> 72.00Improved
Add orthogonality loss (ablation)SGIEL with shape-related and shape-erased objectives · SYSU-MM01 all search, single-shotRank-1: 72.00 -> 73.52Improved
Add gradient-based reweighting of shape objectives (ablation)SGIEL · SYSU-MM01 all search, single-shotRank-1: 73.52 -> 75.18Improved
Replace one orthogonal projector with two unconstrained projectors (ablation)SGIEL · SYSU-MM01 all search, single-shotRank-1: 75.18 -> 70.18Hurt
Remove orthogonality loss while keeping decomposition (ablation)SGIEL · SYSU-MM01 all search, single-shotRank-1: 75.18 -> 73.49Hurt
Add orthogonality loss between two independent projectors (ablation)Two-projector variant · SYSU-MM01 all search, single-shotRank-1: 70.18 -> 72.76Improved
Use one projector rather than two independent projectors, without orthogonality loss (ablation)Shape-erased feature learning · SYSU-MM01 all search, single-shotRank-1: 70.18 -> 73.49Improved
Concatenate outputs of two checkpointsOurs · SYSU-MM01 all search, single-shotRank-1: 75.18 -> 77.12Improved
Use SGIEL rather than the classification-and-triplet baselineHITSZ-VCM · HITSZ-VCM infrared to visibleRank-1: 62.02 -> 67.65Improved
Use SGIEL rather than the classification-and-triplet baselineHITSZ-VCM · HITSZ-VCM visible to infraredRank-1: 64.90 -> 70.23Improved

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