AI Paper Atlas

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

Channel Augmented Joint Learning for Visible-Infrared Recognition

Mang Ye, Weijian Ruan, Bo Du, Mike Zheng Shou

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Margin-based face recognition losses · 413 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 random channel exchangeable augmentation (ablation)AGW baseline · SYSU-MM01 all searchRank-1: 49.87 -> 54.29Improved
add channel-level random erasing to channel augmentation (ablation)AGW + CA · SYSU-MM01 all searchRank-1: 54.29 -> 65.28Improved
add random grayscale augmentation (ablation)AGW + CA + CRE · SYSU-MM01 all searchRank-1: 65.28 -> 65.91Improved
add random horizontal flip (ablation)AGW + CA + CRE + GA · SYSU-MM01 all searchRank-1: 65.91 -> 67.02Improved
add enhanced squared difference to channel-mixed learning (ablation)M · SYSU-MM01 all searchRank-1: 67.02 -> 69.16Improved
add enhanced squared difference to joint learning (ablation)J · SYSU-MM01 all searchRank-1: 67.68 -> 69.88Improved
add channel-augmented joint learning (with squared difference) (ablation)M · SYSU-MM01 all searchRank-1: 69.16 -> 69.88Improved
add channel-augmented joint learning (ablation)AGW + augmentations · SYSU-MM01 all searchRank-1: 67.02 -> 69.88Improved
use squared difference instead of hard triplet mining (ablation)J · SYSU-MM01 all searchRank-1: 66.41 -> 69.88Improved
add channel augmentation and ours to DDAGDDAG · SYSU-MM01 all searchRank-1: 54.75 -> 62.51Improved
add channel augmentation and random erasing to LightCNN-29LightCNN-29 · Oulu-CASIA NIR-VISVR@FAR=0.1%: 87.0 -> 91.7Improved
add channel augmentation and random erasing to LightCNN-29LightCNN-29 · BUAA-VisNirRank-1: 98.0 -> 98.3Improved

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