Compact Occlusion-Robust Facial Expression Recognition via Clean-Anchored Hard Occlusion Fine-Tuning
Sensors · 2026
DOI: 10.3390/s26175500 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, 2026. Verfasst von Xuefeng Zhao und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer Engineering, Jiangsu Ocean University, Lianyungang 222005, China. Hauptthemen laut MEDLINE: Facial Expression, Automated Facial Recognition, Image Processing, Computer-Assisted, Pattern Recognition, Automated. Die Arbeit nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Jiangsu Provincial Marine Science and Technology Innovation Project, Jiangsu Provincial Water Conservancy Science and Technology Project, Lianyungang "521 Project" Research Funding Program.
Abstract (Originalsprache)
Facial occlusion removes expression-relevant evidence and remains a major source of error in camera-based affective sensing. Existing approaches to occlusion-robust facial expression recognition often rely on specialized attention, reconstruction, semantic, or geometric pipelines, whereas aggressive training using synthetic occlusions may impair discrimination on clean images or overfit to synthetic corruption patterns. We address this tension between cleanness and robustness through clean-anchored hard occlusion fine-tuning (CA-HOFT). HardMix samples structured and random occlusion modes according to a facial region-weighted distribution. An explicit classification branch for the non-HardMix source view preserves ground-truth supervision, while a fixed reference teacher initialized from the preceding mixed-occlusion stage supplies a stationary distribution for both paired student views. These signals jointly train a single classifier while retaining a single-backbone inference pathway. Across five independent training runs, the ResNet-18 student achieves 90.08% accuracy on the Real-World Affective Faces Database (RAF-DB) and 87.38% accuracy with 83.34% macro-F1 on Occlusion-RAF-DB, with corresponding sample standard deviations of 0.21, 0.12, and 0.32 percentage points. The retained model has 11.18 million parameters and requires 1.814 giga multiply–accumulate operations (GMACs). Controlled comparisons and ablations indicate that it has the most favorable observed cleanness–robustness trade-off among the tested epoch-matched alternatives; however, fixed-checkpoint comparisons on Occlusion-RAF-DB are not significant after Holm correction. AffectNet-8 and evaluations using natural occlusion provide supporting evidence, while broader cross-domain validation remains future work.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Computer Engineering, Jiangsu Ocean University, Lianyungang 222005, China.
- School of Smart Ocean, Jiangsu Ocean University, Lianyungang 222005, China.
- School of Marine Technology and Geomatics, Jiangsu Ocean University, Lianyungang 222005, China.
ORCID-Profile: Xuefeng Zhao
Förderung
- Jiangsu Provincial Marine Science and Technology Innovation Project
- Jiangsu Provincial Water Conservancy Science and Technology Project
- Lianyungang "521 Project" Research Funding Program
- Lianyungang Key Research and Development Program (Industrial Foresight and Key Technologies)
- Lianyungang “521 Project” Research Funding Program
Zitieren
APA
Zhao, X., Dong, Y., Zhong, Z., & Jiang, T. (2026). Compact Occlusion-Robust Facial Expression Recognition via Clean-Anchored Hard Occlusion Fine-Tuning. Sensors. https://doi.org/10.3390/s26175500
Vancouver
Zhao X, Dong Y, Zhong Z, Jiang T. Compact Occlusion-Robust Facial Expression Recognition via Clean-Anchored Hard Occlusion Fine-Tuning. Sensors. 2026. doi:10.3390/s26175500
BibTeX
@article{Zhao2026,
author = {Zhao, Xuefeng and Dong, Yixuan and Zhong, Zhaoman and Jiang, Tingchen},
title = {Compact Occlusion-Robust Facial Expression Recognition via Clean-Anchored Hard Occlusion Fine-Tuning},
journal = {Sensors},
year = {2026},
doi = {10.3390/s26175500}
}
EUVIMED