Occlusion-Aware Face Recognition via Adaptive Local Feature Fusion and Identity-Guided Contrastive Learning
Kexin Zhu, Guoqing Ma · 2026
Read Occlusion-Aware Face Recognition via Adaptive Local Feature Fusion and Identity-Guided Contrastive Learning by Kexin Zhu, Guoqing Ma (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13364154.txt.
Description
Kexin Zhu and Guoqing Ma present a 2026 study on face recognition when part of the face is covered by masks, hats, or sunglasses. Covered regions hide identity cues and inject noise. So their framework scores the trustworthiness of three face zones and fuses the local features with those weights. It trains with a joint loss that pulls same-person features together while pushing classes apart.
Across four covering types, the method reached about 92 percent recognition accuracy, beating three strong rival systems by clear margins on both correct accepts and error rates. The design is kept light for practical use. The authors present region reliability as the key move: lean on what the camera can truly see.
Catalog record
| CREATOR | Kexin Zhu, Guoqing Ma |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52482 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |