AI Paper Atlas

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

Towards Interpretable Face Recognition

Bangjie Yin, Luan Tran, Haoxiang Li, Xiaohui Shen, Xiaoming Liu

ICCV 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Margin-based face recognition losses · 96 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
use FAD threshold t=260 instead of forcing all elements equal (t=320) (ablation)CASIA-Net with SAD+FAD · IJB-Averification TAR @FAR=.01 (%): 74.6 -> 79.2Improved
use FAD threshold t=260 instead of t=130 (ablation)CASIA-Net with SAD+FAD · IJB-Averification TAR @FAR=.01 (%): 79.0 -> 79.2No clear effect
add FAD loss to SAD loss (ablation)CASIA-Net · IJB-A manual occlusionverification TAR @FAR=.01 (%): 66.6 -> 76.1Improved
add FAD loss to SAD loss (ablation)CASIA-Net · IJB-A manual occlusionidentification Rank-1 (%): 81.2 -> 88.0Improved
add FAD loss to SAD loss (ablation)CASIA-Net · IJB-Averification TAR @FAR=.01 (%): 78.1 -> 79.2Improved
add SAD loss to FAD loss (ablation)CASIA-Net · IJB-Averification TAR @FAR=.01 (%): 76.7 -> 79.2Improved
add SAD loss to FAD loss (ablation)CASIA-Net · IJB-A manual occlusionidentification Rank-1 (%): 85.1 -> 88.0Improved
replace black static window with Gaussian noise dynamic-size occlusion in FAD (ablation)CASIA-Net · IJB-Aidentification Rank-1 (%): 89.4 -> 89.9Improved
add large magnitude filtering with d=95.83% (ablation)CASIA-Net · peak location std (pos./neg.)std of peak locations: 25.7/25.7 -> 12.9/13.4Improved
add SAD loss to base CNN (ablation)CASIA-Net · spreadness of average peak locationsspreadness d-bar: 6.9 -> 17.1Improved
add FAD loss to SAD only (ablation)CASIA-Net · spreadness of average peak locationsspreadness d-bar: 17.1 -> 18.7Improved
train baseline CASIA-Net with synthetic occlusion augmentationCASIA-Net · IJB-Averification TAR @FAR=.01 (%): 74.3 -> 78.9Improved

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