Keywords
Uncertainty
Sample selection strategy
Submitted : 2025-03-04
Accepted : 2025-03-19
Published : 2025-04-03
Abstract
In the task of Facial Expression Recognition (FER), data uncertainty has been a critical factor affecting performance, typically arising from the ambiguity of facial expressions, low-quality images, and the subjectivity of annotators. Tracking the training history reveals that misclassified samples often exhibit high confidence and excessive uncertainty in the early stages of training. To address this issue, we propose an uncertainty-based robust sample selection strategy, which combines confidence error with RandAugment to improve image diversity, effectively reducing overfitting caused by uncertain samples during deep learning model training. To validate the effectiveness of the proposed method, extensive experiments were conducted on FER public benchmarks. The accuracy obtained were 89.08% on RAF-DB, 63.12% on AffectNet, and 88.73% on FERPlus.
References
Wang K, Peng X, Yang J, et al., 2020, Suppressing Uncertainties for Large-scale Facial Expression Recognition. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 6897–6906.
She J, Hu Y, Shi H, et al., 2021, Dive Into Ambiguity: Latent Distribution Mining and Pairwise Uncertainty Estimation for Facial Expression Recognition. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 6248–6257.
Zhang Y, Wang C, Deng W, 2021, Relative Uncertainty Learning for Facial Expression Recognition. Advances in Neural Information Processing Systems, 34: 17616–17627.
Zeng J, Shan S, Chen X, 2018, Facial Expression Recognition with Inconsistently Annotated Datasets. Proceedings of the Proceedings of the European Conference on Computer Vision (ECCV), 222–237.
Cubuk ED, Zoph B, Shlens J, et al., 2020, Randaugment: Practical Automated Data Augmentation with a Reduced Search Space. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 702–703.
Yao Y, Liu T, Gong M, et al., 2021, Instance-dependent Label-noise Learning Under a Structural Causal Model. Advances in Neural Information Processing Systems, 34: 4409–4420.
Yao Y, Liu T, Han B, et al., 2020, Dual t: Reducing Estimation Error for Transition Matrix in Label-noise Learning. Advances in Neural Information Processing Systems, 33: 7260–7271.
Nguyen D, Mummadi C, Ngo T, et al., 2019, Self: Learning to Filter Noisy Labels with Self-ensembling. arXiv: 1910.01842. https://doi.org/10.48550/arXiv.1910.01842.
Torkzadehmahani R, Nasirigerdeh R, Rueckert D, et al., 2022, Label Noise-robust Learning using a Confidence-based Sieving Strategy. arXiv: 2210.05330. https://doi.org/10.48550/arXiv.2210.05330.
Li S, Deng W, Du J, 2017, Reliable Crowdsourcing and Deep Locality-preserving Learning for Expression Recognition in the Wild. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2852–2861.
Barsoum E, Zhang C, Ferrer C, et al., 2016, Training Deep Networks for Facial Expression Recognition with Crowd-sourced Label Distribution. Proceedings of the Proceedings of the 18th ACM International Conference on Multimodal Interaction, 279–283.
Mollahosseini A, Hasani B, Mahoor M, 2017, Affectnet: A Database for Facial Expression, Valence, and Arousal Computing in the Wild. IEEE Transactions on Affective Computing, 10(1): 18–31.
He K, Zhang X, Ren S, et al., 2016, Deep Residual Learning for Image Recognition, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 770–778.
Guo Y, Zhang L, Hu Y, et al., 2016, Ms-celeb-1m: A dataset and Benchmark for Large-scale Face Recognition. Proceedings of the Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part III 14, 2016, Springer International Publishing, 87–102.
Zhong Z, Zheng L, Kang G, et al., 2020, Random Erasing Data Augmentation. Proceedings of the AAAI Conference on Artificial Intelligence, 34(07): 13001–13008.