{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,31]],"date-time":"2026-08-31T06:17:28Z","timestamp":1788157048611,"version":"build-2803163510"},"reference-count":41,"publisher":"MDPI AG","issue":"17","license":[{"start":{"date-parts":[[2026,8,30]],"date-time":"2026-08-30T00:00:00Z","timestamp":1788048000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Jiangsu Provincial Marine Science and Technology Innovation Project","award":["JSZRHYKJ202201"],"award-info":[{"award-number":["JSZRHYKJ202201"]}]},{"name":"Jiangsu Provincial Water Conservancy Science and Technology Project","award":["2020058"],"award-info":[{"award-number":["2020058"]}]},{"name":"Lianyungang \u201c521 Project\u201d Research Funding Program","award":["LYG06521202131"],"award-info":[{"award-number":["LYG06521202131"]}]},{"name":"Lianyungang Key Research and Development Program","award":["CG2527"],"award-info":[{"award-number":["CG2527"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>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\u2013accumulate operations (GMACs). Controlled comparisons and ablations indicate that it has the most favorable observed cleanness\u2013robustness 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.<\/jats:p>","DOI":"10.3390\/s26175500","type":"journal-article","created":{"date-parts":[[2026,8,31]],"date-time":"2026-08-31T05:55:56Z","timestamp":1788155756000},"page":"5500","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Compact Occlusion-Robust Facial Expression Recognition via Clean-Anchored Hard Occlusion Fine-Tuning"],"prefix":"10.3390","volume":"26","author":[{"ORCID":"https:\/\/orcid.org\/0009-0004-0628-4320","authenticated-orcid":false,"given":"Xuefeng","family":"Zhao","sequence":"first","affiliation":[{"name":"School of Computer Engineering, Jiangsu Ocean University, Lianyungang 222005, China"},{"name":"School of Smart Ocean, Jiangsu Ocean University, Lianyungang 222005, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"conceptualization","vocabulary":"credit"},{"role":"methodology","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"},{"role":"supervision","vocabulary":"credit"},{"role":"project-administration","vocabulary":"credit"}]},{"given":"Yixuan","family":"Dong","sequence":"additional","affiliation":[{"name":"School of Computer Engineering, Jiangsu Ocean University, Lianyungang 222005, China"},{"name":"School of Smart Ocean, Jiangsu Ocean University, Lianyungang 222005, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"software","vocabulary":"credit"},{"role":"validation","vocabulary":"credit"},{"role":"formal-analysis","vocabulary":"credit"},{"role":"investigation","vocabulary":"credit"},{"role":"visualization","vocabulary":"credit"},{"role":"data-curation","vocabulary":"credit"},{"role":"writing-original-draft","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"}]},{"given":"Zhaoman","family":"Zhong","sequence":"additional","affiliation":[{"name":"School of Computer Engineering, Jiangsu Ocean University, Lianyungang 222005, China"},{"name":"School of Smart Ocean, Jiangsu Ocean University, Lianyungang 222005, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"methodology","vocabulary":"credit"},{"role":"validation","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"},{"role":"funding-acquisition","vocabulary":"credit"}]},{"given":"Tingchen","family":"Jiang","sequence":"additional","affiliation":[{"name":"School of Marine Technology and Geomatics, Jiangsu Ocean University, Lianyungang 222005, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"validation","vocabulary":"credit"},{"role":"resources","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"},{"role":"funding-acquisition","vocabulary":"credit"}]}],"member":"1968","published-online":{"date-parts":[[2026,8,30]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"4057","DOI":"10.1109\/TIP.2019.2956143","article-title":"Region attention networks for pose and occlusion robust facial expression recognition","volume":"29","author":"Wang","year":"2020","journal-title":"IEEE Trans. Image Process."},{"key":"ref_2","unstructured":"Ding, H., Zhou, P., and Chellappa, R. (October, January 28). Occlusion-adaptive deep network for robust facial expression recognition. Proceedings of the IEEE International Joint Conference on Biometrics (IJCB), Houston, TX, USA."},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Lee, I., Lee, E., and Yoo, S.B. (2023, January 1\u20136). Latent-OFER: Detect, mask, and reconstruct with latent vectors for occluded facial expression recognition. Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV), Paris, France.","DOI":"10.1109\/ICCV51070.2023.00148"},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"Zhai, H., Yang, X., Ye, Y., Li, C., Fan, B., and Li, C. (2025, January 27\u201331). Rethinking occlusion in FER: A semantic-aware perspective and go beyond. Proceedings of the 33rd ACM International Conference on Multimedia, Dublin, Ireland.","DOI":"10.1145\/3746027.3754978"},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Mukhiddinov, M., Djuraev, O., Akhmedov, F., Mukhamadiyev, A., and Cho, J. (2023). Masked face emotion recognition based on facial landmarks and deep learning approaches for visually impaired people. Sensors, 23.","DOI":"10.3390\/s23031080"},{"key":"ref_6","doi-asserted-by":"crossref","unstructured":"So, J., and Han, Y. (2025). Facial landmark-driven keypoint feature extraction for robust facial expression recognition. Sensors, 25.","DOI":"10.3390\/s25123762"},{"key":"ref_7","doi-asserted-by":"crossref","unstructured":"Wang, K., Peng, X., Yang, J., Lu, S., and Qiao, Y. (2020, January 13\u201319). Suppressing uncertainties for large-scale facial expression recognition. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA.","DOI":"10.1109\/CVPR42600.2020.00693"},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Zeng, J., Shan, S., and Chen, X. (2018). Facial expression recognition with inconsistently annotated datasets. Computer Vision\u2014ECCV 2018, Springer.","DOI":"10.1007\/978-3-030-01261-8_14"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"2531","DOI":"10.1109\/TMM.2023.3301209","article-title":"Uncertain facial expression recognition via multi-task assisted correction","volume":"26","author":"Liu","year":"2024","journal-title":"IEEE Trans. Multimed."},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Farzaneh, A.H., and Qi, X. (2021, January 3\u20138). Facial expression recognition in the wild via deep attentive center loss. Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV), Waikoloa, HI, USA.","DOI":"10.1109\/WACV48630.2021.00245"},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"6535","DOI":"10.1109\/TIP.2020.2991510","article-title":"Semantic neighborhood-aware deep facial expression recognition","volume":"29","author":"Fu","year":"2020","journal-title":"IEEE Trans. Image Process."},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Zheng, C., Mendieta, M., and Chen, C. (2023, January 2\u20136). POSTER: A pyramid cross-fusion transformer network for facial expression recognition. Proceedings of the IEEE\/CVF International Conference on Computer Vision Workshops (ICCVW), Paris, France.","DOI":"10.1109\/ICCVW60793.2023.00339"},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"110951","DOI":"10.1016\/j.patcog.2024.110951","article-title":"POSTER++: A simpler and stronger facial expression recognition network","volume":"157","author":"Mao","year":"2025","journal-title":"Pattern Recognit."},{"key":"ref_14","doi-asserted-by":"crossref","unstructured":"Li, H., Niu, H., Zhu, Z., and Zhao, F. (2024, January 15\u201319). CLIPER: A unified vision\u2013language framework for in-the-wild facial expression recognition. Proceedings of the IEEE International Conference on Multimedia and Expo (ICME), Niagara Falls, ON, Canada.","DOI":"10.1109\/ICME57554.2024.10687508"},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Cabacas-Maso, J., Ortega-Beltr\u00e1n, E., Benito-Altamirano, I., and Ventura, C. (2025, January 11\u201312). Enhancing facial expression recognition with LSTM through dual-direction attention mixed feature networks and CLIP. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Nashville, TN, USA.","DOI":"10.1109\/CVPRW67362.2025.00564"},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"He, X., Luo, C., Xian, X., Li, B., Khan, M.H., Ge, Z., Xie, W., Song, S., Shen, L., and Ghanem, B. (2025, January 19\u201325). SynFER: Towards boosting facial expression recognition with synthetic data. Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV), Honolulu, HI, USA.","DOI":"10.1109\/ICCV51701.2025.00949"},{"key":"ref_17","unstructured":"Zhao, Z., Liu, Q., and Zhou, F. (2021, January 2\u20139). Robust lightweight facial expression recognition network with label distribution training. Proceedings of the AAAI Conference on Artificial Intelligence, Virtual Event."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Guo, Y., Xia, Y., Wang, J., Yu, H., and Chen, R.-C. (2020). Real-time facial affective computing on mobile devices. Sensors, 20.","DOI":"10.3390\/s20030870"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Pascual, A.M., Valverde, E.C., Kim, J.-I., Jeong, J.-W., Jung, Y., Kim, S.-H., and Lim, W. (2022). Light-FER: A lightweight facial emotion recognition system on edge devices. Sensors, 22.","DOI":"10.3390\/s22239524"},{"key":"ref_20","doi-asserted-by":"crossref","unstructured":"Chen, X., and Huang, L. (2024). A lightweight model enhancing facial expression recognition with spatial bias and cosine-harmony loss. Computation, 12.","DOI":"10.20944\/preprints202408.1304.v1"},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"2439","DOI":"10.1109\/TIP.2018.2886767","article-title":"Occlusion aware facial expression recognition using CNN with attention mechanism","volume":"28","author":"Li","year":"2019","journal-title":"IEEE Trans. Image Process."},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Kim, J., and Lee, D. (2023). Facial expression recognition robust to occlusion and to intra-similarity problem using relevant subsampling. Sensors, 23.","DOI":"10.3390\/s23052619"},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Wang, J., Ding, H., and Wang, S. (2023). Occluded facial expression recognition using self-supervised learning. Computer Vision\u2014ACCV 2022, Springer.","DOI":"10.1007\/978-3-031-26316-3_8"},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"Ortmann, T., Wang, Q., and Putzar, L. (2024, January 15\u201318). EmojiHeroVR: A study on facial expression recognition under partial occlusion from head-mounted displays. Proceedings of the 12th International Conference on Affective Computing and Intelligent Interaction (ACII), Glasgow, UK.","DOI":"10.1109\/ACII63134.2024.00014"},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Ortmann, T., Wang, Q., and Putzar, L. (2025, January 27\u201329). Unimodal and multimodal static facial expression recognition for virtual reality users with EmoHeVRDB. Proceedings of the IEEE International Conference on Artificial Intelligence and eXtended and Virtual Reality (AIxVR), Lisbon, Portugal.","DOI":"10.1109\/AIxVR63409.2025.00048"},{"key":"ref_26","unstructured":"Zhong, Z., Zheng, L., Kang, G., Li, S., and Yang, Y. (2020, January 7\u201312). Random erasing data augmentation. Proceedings of the AAAI Conference on Artificial Intelligence, New York, NY, USA."},{"key":"ref_27","unstructured":"DeVries, T., and Taylor, G.W. (2017). Improved regularization of convolutional neural networks with cutout. arXiv."},{"key":"ref_28","unstructured":"Yun, S., Han, D., Oh, S.J., Chun, S., Choe, J., and Yoo, Y. (November, January 27). CutMix: Regularization strategy to train strong classifiers with localizable features. Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV), Seoul, Republic of Korea."},{"key":"ref_29","doi-asserted-by":"crossref","unstructured":"Zhang, Y., Wang, C., Ling, X., and Deng, W. (2022). Learn from all: Erasing attention consistency for noisy label facial expression recognition. Computer Vision\u2014ECCV 2022, Springer.","DOI":"10.1007\/978-3-031-19809-0_24"},{"key":"ref_30","unstructured":"Hinton, G., Vinyals, O., and Dean, J. (2015). Distilling the knowledge in a neural network. arXiv."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"1789","DOI":"10.1007\/s11263-021-01453-z","article-title":"Knowledge distillation: A survey","volume":"129","author":"Gou","year":"2021","journal-title":"Int. J. Comput. Vis."},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"2935","DOI":"10.1109\/TPAMI.2017.2773081","article-title":"Learning without forgetting","volume":"40","author":"Li","year":"2018","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_33","unstructured":"Furlanello, T., Lipton, Z.C., Tschannen, M., Itti, L., and Anandkumar, A. (2018, January 10\u201315). Born again neural networks. Proceedings of the 35th International Conference on Machine Learning (ICML), Stockholm, Sweden. Available online: https:\/\/proceedings.mlr.press\/v80\/furlanello18a.html."},{"key":"ref_34","first-page":"1195","article-title":"Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results","volume":"Volume 30","author":"Tarvainen","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref_35","first-page":"6256","article-title":"Unsupervised data augmentation for consistency training","volume":"Volume 33","author":"Xie","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref_36","doi-asserted-by":"crossref","unstructured":"Xie, Q., Luong, M.-T., Hovy, E., and Le, Q.V. (2020, January 13\u201319). Self-training with noisy student improves ImageNet classification. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA.","DOI":"10.1109\/CVPR42600.2020.01070"},{"key":"ref_37","first-page":"596","article-title":"FixMatch: Simplifying semi-supervised learning with consistency and confidence","volume":"Volume 33","author":"Sohn","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref_38","doi-asserted-by":"crossref","unstructured":"Li, S., Deng, W., and Du, J. (2017, January 21\u201326). Reliable crowdsourcing and deep locality-preserving learning for expression recognition in the wild. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA.","DOI":"10.1109\/CVPR.2017.277"},{"key":"ref_39","doi-asserted-by":"crossref","first-page":"18","DOI":"10.1109\/TAFFC.2017.2740923","article-title":"AffectNet: A database for facial expression, valence, and arousal computing in the wild","volume":"10","author":"Mollahosseini","year":"2019","journal-title":"IEEE Trans. Affect. Comput."},{"key":"ref_40","doi-asserted-by":"crossref","unstructured":"He, K., Zhang, X., Ren, S., and Sun, J. (2016, January 27\u201330). Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA.","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref_41","doi-asserted-by":"crossref","unstructured":"Guo, Y., Zhang, L., Hu, Y., He, X., and Gao, J. (2016). MS-Celeb-1M: A dataset and benchmark for large-scale face recognition. Computer Vision\u2014ECCV 2016, Springer.","DOI":"10.1007\/978-3-319-46487-9_6"}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/17\/5500\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,31]],"date-time":"2026-08-31T06:00:37Z","timestamp":1788156037000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/17\/5500"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,30]]},"references-count":41,"journal-issue":{"issue":"17","published-online":{"date-parts":[[2026,9]]}},"alternative-id":["s26175500"],"URL":"https:\/\/doi.org\/10.3390\/s26175500","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,30]]}}}