{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T02:28:18Z","timestamp":1778725698571,"version":"3.51.4"},"reference-count":57,"publisher":"MDPI AG","issue":"5","license":[{"start":{"date-parts":[[2026,4,28]],"date-time":"2026-04-28T00:00:00Z","timestamp":1777334400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["MAKE"],"abstract":"<jats:p>Online and blended classrooms widen access but remove the in-person cues instructors use to gauge attention. Prior work typically relies on heavy, cloud-bound or multimodal models that are hard to deploy on commodity laptops, treats attention as an unordered label without calibrated probabilities, and evaluates on subject-overlapping splits with limited robustness analysis. This creates a gap in Tiny, deployable, calibration-aware methods validated under realistic protocols. We address this gap with a TinyML, vision-only pipeline that estimates four attention levels: (Very Low, low, high, Very High ) from short webcam clips under strict on-device budgets. Each clip of T=30 frames at 224\u00d7224 is processed by a compact hybrid encoder: a CNN extracts per frame spatial features, a BiLSTM models temporal context, and a lightweight GRU refines dynamics; three parallel branches with staggered widths encourage feature diversity before fusion. We apply structured pruning of convolutional channels and recurrent units, post-training INT8 quantization, and temperature scaling for calibrated probabilities; models are exported as ONNX. On DAiSEE with subject-independent splits, the baseline attains 99.86% accuracy and 0.998 macro-F1, with strong ordinal agreement (QWK = 0.998, ordinal MAE = 0.03). The compressed model preserves reliability (macro-F1 = 0.995, QWK = 0.995), remains robust to low light, partial occlusion, and head yaw, and yields \u223c4\u00d7 smaller size and \u223c2.3\u00d7 CPU speedups. These results indicate a deployable, privacy-preserving approach to fine-grained, on-device attention analytics.<\/jats:p>","DOI":"10.3390\/make8050116","type":"journal-article","created":{"date-parts":[[2026,4,28]],"date-time":"2026-04-28T15:09:33Z","timestamp":1777388973000},"page":"116","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["A Tiny Vision-Based Model for Real-Time Student Attention Detection in Online Classes"],"prefix":"10.3390","volume":"8","author":[{"ORCID":"https:\/\/orcid.org\/0009-0002-2316-8140","authenticated-orcid":false,"given":"Chaymae","family":"Yahyati","sequence":"first","affiliation":[{"name":"Multidisciplinary Faculty of Nador, Mohammed Premier University, Oujda 60000, Morocco"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-6532-2573","authenticated-orcid":false,"given":"Ismail","family":"Lamaakal","sequence":"additional","affiliation":[{"name":"Multidisciplinary Faculty of Nador, Mohammed Premier University, Oujda 60000, Morocco"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4704-5364","authenticated-orcid":false,"given":"Yassine","family":"Maleh","sequence":"additional","affiliation":[{"name":"Laboratory LaSTI, ENSAK, Sultan Moulay Slimane University, Khouribga 23000, Morocco"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9569-9162","authenticated-orcid":false,"given":"Khalid","family":"El Makkaoui","sequence":"additional","affiliation":[{"name":"Multidisciplinary Faculty of Nador, Mohammed Premier University, Oujda 60000, Morocco"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0552-9275","authenticated-orcid":false,"given":"Ibrahim","family":"Ouahbi","sequence":"additional","affiliation":[{"name":"Multidisciplinary Faculty of Nador, Mohammed Premier University, Oujda 60000, Morocco"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2026,4,28]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"128379","DOI":"10.1109\/ACCESS.2025.3589938","article-title":"Human behavior analysis: A comprehensive survey on techniques, applications, challenges, and future directions","volume":"13","author":"Essahraui","year":"2025","journal-title":"IEEE Access"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"100394","DOI":"10.1016\/j.edurev.2021.100394","article-title":"Facilitating flexible learning by replacing classroom time with an online learning environment: A systematic review of blended learning in higher education","volume":"34","author":"Mildenberger","year":"2021","journal-title":"Educ. Res. Rev."},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Abdellaoui Alaoui, E.A., Merras, M., and Nayyar, A. (2026). Student Performance Prediction Based on Ensemble Learning Techniques. Innovative Approaches and Applications for Sustainable Development, Springer.","DOI":"10.1007\/978-3-031-97825-8"},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"Lamaakal, I., Maleh, Y., El Makkaoui, K., Ouahbi, I., and Abd El-Latif, A. (2026). Innovating Education 6.0 with LLM Classroom Strategies for Robotics and Computing Skills Integration in EFL Ecosystem. Theory, Practice, and Future Direction of Large Language Models, IGI Global Scientific Publishing.","DOI":"10.4018\/979-8-3693-8387-2"},{"key":"ref_5","first-page":"3151","article-title":"Deep learning models for detecting cheating in online exams","volume":"85","author":"Essahraui","year":"2025","journal-title":"Comput. Mater. Contin."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"87","DOI":"10.1177\/1474022220906393","article-title":"Hands-on versus virtual: Reshaping the design classroom with blended learning","volume":"20","author":"Fleischmann","year":"2021","journal-title":"Arts Humanit. High. Educ."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"100191","DOI":"10.1016\/j.caeai.2023.100191","article-title":"Attention monitoring of students during online classes using XGBoost classifier","volume":"5","author":"Hossen","year":"2023","journal-title":"Comput. Educ. Artif. Intell."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"103779","DOI":"10.1016\/j.compedu.2019.103779","article-title":"Does visual attention to the instructor in online video affect learning and learner perceptions? An eye-tracking analysis","volume":"146","author":"Wang","year":"2020","journal-title":"Comput. Educ."},{"key":"ref_9","first-page":"1012","article-title":"Multimodal engagement analysis from facial videos in the classroom","volume":"14","author":"Goldberg","year":"2021","journal-title":"IEEE Trans. Affect. Comput."},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"4069","DOI":"10.1007\/s10639-022-11370-4","article-title":"Facial emotion recognition of deaf and hard-of-hearing students for engagement detection using deep learning","volume":"28","author":"Lasri","year":"2023","journal-title":"Educ. Inf. Technol."},{"key":"ref_11","first-page":"58","article-title":"Students\u2019 attention in class: Patterns, perceptions of cause and a tool for measuring classroom quality of life","volume":"8","author":"Allison","year":"2020","journal-title":"J. Perspect. Appl. Acad. Pract."},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"32419","DOI":"10.1109\/JIOT.2025.3565688","article-title":"A comprehensive survey on tiny machine learning for human behavior analysis","volume":"12","author":"Lamaakal","year":"2025","journal-title":"IEEE Internet Things J."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"55756","DOI":"10.1109\/JIOT.2025.3625087","article-title":"An explainable Tiny-Fast Kolmogorov\u2013Arnold network for gesture-based air handwriting recognition of Tifinagh letters in resource-constrained IoT device","volume":"12","author":"Lamaakal","year":"2025","journal-title":"IEEE Internet Things J."},{"key":"ref_14","doi-asserted-by":"crossref","unstructured":"Yahyati, C., Lamaakal, I., El Makkaoui, K., Ouahbi, I., and Maleh, Y. (2025). TinyML: Emerging applications and future research directions. Tiny Machine Learning Techniques for Constrained Devices, CRC.","DOI":"10.1201\/9781003544449-10"},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Yahyati, C., Lamaakal, I., Makkaoui, K.E., Ouahbi, I., and Maleh, Y. (2025). TinyML-based facial recognition for embedded systems. 2025 International Conference on Circuit, Systems and Communication (ICCSC), IEEE.","DOI":"10.1109\/ICCSC66714.2025.11134957"},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Yahyati, C., Lamaakal, I., Makkaoui, K.E., Maleh, Y., and Ouahbi, I. (2025). A survey on TinyML applications in education. 2025 International Conference on Electrical Systems & Automation (ICESA), IEEE.","DOI":"10.1109\/ICESA66763.2025.11280838"},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"66","DOI":"10.1007\/s10648-025-10046-y","article-title":"A systematic review on student engagement in undergraduate mathematics: Conceptualization, measurement, and learning outcomes","volume":"37","author":"Papageorgiou","year":"2025","journal-title":"Educ. Psychol. Rev."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"7143","DOI":"10.1007\/s10639-024-12989-1","article-title":"Student engagement and speaking performance in AI-assisted learning environments: A mixed-methods study from Chinese middle schools","volume":"30","author":"Huang","year":"2025","journal-title":"Educ. Inf. Technol."},{"key":"ref_19","first-page":"3474","article-title":"Multimodal learning engagement assessment system: An innovative approach to optimizing learning engagement","volume":"41","author":"Li","year":"2025","journal-title":"Int. J. Hum. Comput. Interact."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"243","DOI":"10.1109\/TLT.2025.3530457","article-title":"Msc-trans: A multi-feature-fusion network with encoding structure for student engagement detecting","volume":"18","author":"Xie","year":"2025","journal-title":"IEEE Trans. Learn. Technol."},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"2552","DOI":"10.1038\/s41598-025-32521-w","article-title":"Deep learning-based model for analyzing student engagement in activities","volume":"16","author":"Feng","year":"2026","journal-title":"Sci. Rep."},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"69236","DOI":"10.1109\/ACCESS.2025.3561459","article-title":"A Multi-Faceted Deep Learning Approach for Student Engagement Insights and Adaptive Content Recommendations","volume":"13","author":"Saleem","year":"2025","journal-title":"IEEE Access"},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"9989","DOI":"10.1007\/s10639-023-12058-z","article-title":"CNN-Transformer: A deep learning method for automatically identifying learning engagement","volume":"29","author":"Xiong","year":"2024","journal-title":"Educ. Inf. Technol."},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"13","DOI":"10.1007\/s10956-024-10143-7","article-title":"Facial expression recognition for probing students\u2019 emotional engagement in science learning","volume":"34","author":"Tang","year":"2025","journal-title":"J. Sci. Educ. Technol."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"12167","DOI":"10.1109\/ACCESS.2026.3656854","article-title":"A novel FastKAN with few-shot learning for real-time driver distraction detection on TinyML microcontrollers","volume":"14","author":"Yahyati","year":"2026","journal-title":"IEEE Access"},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"52621","DOI":"10.1007\/s11042-023-17533-w","article-title":"A vision-based multi-cues approach for individual students\u2019 and overall class engagement monitoring in smart classroom environments","volume":"83","author":"Pabba","year":"2024","journal-title":"Multimed. Tools Appl."},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"25228","DOI":"10.1109\/ACCESS.2024.3353053","article-title":"Multimodal engagement recognition from image traits using deep learning techniques","volume":"12","author":"Sukumaran","year":"2024","journal-title":"IEEE Access"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"128239","DOI":"10.1016\/j.eswa.2025.128239","article-title":"Video-based real-time monitoring of engagement in E-learning using MediaPipe through multi-feature analysis","volume":"288","author":"Wang","year":"2025","journal-title":"Expert Syst. Appl."},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"37","DOI":"10.1007\/s13278-025-01408-z","article-title":"Development of a framework using deep learning for the identification and classification of engagement levels in distance learning students","volume":"15","author":"Ferreira","year":"2025","journal-title":"Soc. Netw. Anal. Min."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"18","DOI":"10.1007\/s13735-023-00284-7","article-title":"Recognition of student engagement in classroom from affective states","volume":"12","author":"Mandia","year":"2023","journal-title":"Int. J. Multimed. Inf. Retr."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"126426","DOI":"10.1016\/j.eswa.2025.126426","article-title":"Measuring student attention based on EEG brain signals using deep reinforcement learning","volume":"269","author":"Rehman","year":"2025","journal-title":"Expert Syst. Appl."},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"101889","DOI":"10.1016\/j.iot.2026.101889","article-title":"TinyHAR-UQ: Battery-aware, uncertainty-controlled TinyML for wearable activity recognition on IoT edge devices","volume":"36","author":"Lamaakal","year":"2026","journal-title":"Internet Things"},{"key":"ref_33","doi-asserted-by":"crossref","unstructured":"Yahyati, C., Lamaakal, I., El Makkaoui, K., Maleh, Y., and Ouahbi, I. (2025, January 23\u201324). TinyML-IoE: A Unified Architecture for Integrating TinyML into the Internet of Everythings. Proceedings of the 2025 International Conference on Electrical Systems & Automation (ICESA), Troyes, France.","DOI":"10.1109\/ICESA66763.2025.11281241"},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"10564","DOI":"10.1109\/JIOT.2026.3656932","article-title":"A Comprehensive Survey of TinyML-Based Biometric Recognition for IoT Edge Devices","volume":"13","author":"Essahraui","year":"2026","journal-title":"IEEE Internet Things J."},{"key":"ref_35","doi-asserted-by":"crossref","unstructured":"Zhang, W.L., Jia, R.S., Wang, H., Che, C.Y., and Sun, H.M. (2024). A self-supervised learning network for student engagement recognition from facial expressions. IEEE Trans. Circuits Syst. Video Technol., advance online publication.","DOI":"10.1109\/TCSVT.2024.3436029"},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"43","DOI":"10.1007\/s13278-023-01181-x","article-title":"Online learners\u2019 engagement detection via facial emotion recognition in online learning context using hybrid classification model","volume":"14","author":"Maddu","year":"2024","journal-title":"Soc. Netw. Anal. Min."},{"key":"ref_37","doi-asserted-by":"crossref","first-page":"12575","DOI":"10.1007\/s11042-024-19392-5","article-title":"Revolutionizing online education: Advanced facial expression recognition for real-time student progress tracking via deep learning model","volume":"84","author":"Aly","year":"2024","journal-title":"Multimed. Tools Appl."},{"key":"ref_38","doi-asserted-by":"crossref","first-page":"66110","DOI":"10.1109\/ACCESS.2024.3354981","article-title":"Student recognition and activity monitoring in e-classes using deep learning in higher education","volume":"12","author":"Alruwais","year":"2024","journal-title":"IEEE Access"},{"key":"ref_39","unstructured":"Gupta, A., D\u2019Cunha, A., Awasthi, K., and Balasubramanian, V. (2016). Daisee: Towards user engagement recognition in the wild. arXiv."},{"key":"ref_40","doi-asserted-by":"crossref","first-page":"3582","DOI":"10.1109\/TIP.2019.2963380","article-title":"Pwstablenet: Learning pixel-wise warping maps for video stabilization","volume":"29","author":"Zhao","year":"2020","journal-title":"IEEE Trans. Image Process."},{"key":"ref_41","doi-asserted-by":"crossref","first-page":"112086","DOI":"10.1016\/j.matdes.2023.112086","article-title":"Robustness of machine learning to color, size change, normalization, and image enhancement on micrograph datasets with large sample differences","volume":"232","author":"Pei","year":"2023","journal-title":"Mater. Des."},{"key":"ref_42","first-page":"9225","article-title":"Bnet: Batch normalization with enhanced linear transformation","volume":"45","author":"Xu","year":"2023","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_43","doi-asserted-by":"crossref","first-page":"18","DOI":"10.1111\/itor.13471","article-title":"Comparing QUBO models for quantum annealing: Integer encodings for permutation problems","volume":"32","author":"Codognet","year":"2025","journal-title":"Int. Trans. Oper. Res."},{"key":"ref_44","doi-asserted-by":"crossref","first-page":"1905","DOI":"10.1007\/s10462-022-10213-5","article-title":"A review of convolutional neural network architectures and their optimizations","volume":"56","author":"Cong","year":"2023","journal-title":"Artif. Intell. Rev."},{"key":"ref_45","doi-asserted-by":"crossref","first-page":"1235","DOI":"10.1162\/neco_a_01199","article-title":"A review of recurrent neural networks: LSTM cells and network architectures","volume":"31","author":"Yu","year":"2019","journal-title":"Neural Comput."},{"key":"ref_46","doi-asserted-by":"crossref","first-page":"42335","DOI":"10.1038\/s41598-025-26297-2","article-title":"A tiny inertial transformer for human activity recognition via multimodal knowledge distillation and explainable AI","volume":"15","author":"Lamaakal","year":"2025","journal-title":"Sci. Rep."},{"key":"ref_47","doi-asserted-by":"crossref","first-page":"103462","DOI":"10.1016\/j.mex.2025.103462","article-title":"Performance analysis of neural network architectures for time series forecasting: A comparative study of RNN, LSTM, GRU, and hybrid models","volume":"15","author":"Yunita","year":"2025","journal-title":"MethodsX"},{"key":"ref_48","doi-asserted-by":"crossref","unstructured":"Lamaakal, I., Maleh, Y., El Makkaoui, K., Ouahbi, I., P\u0142awiak, P., Alfarraj, O., and Abd El-Latif, A.A. (2025). Tiny language models for automation and control: Overview, potential applications, and future research directions. Sensors, 25.","DOI":"10.3390\/s25051318"},{"key":"ref_49","doi-asserted-by":"crossref","first-page":"204513","DOI":"10.1109\/ACCESS.2025.3633575","article-title":"A systematic review of state-of-the-art TinyML applications in healthcare, education, and transportation","volume":"13","author":"Yahyati","year":"2025","journal-title":"IEEE Access"},{"key":"ref_50","doi-asserted-by":"crossref","unstructured":"Ren, D., Li, W., Ding, T., Wang, L., Fan, Q., Huo, J., and Gao, Y. (2025). Onnxpruner: Onnx-based general model pruning adapter. IEEE Trans. Pattern Anal. Mach. Intell., advance online publication.","DOI":"10.1109\/TPAMI.2025.3554560"},{"key":"ref_51","doi-asserted-by":"crossref","unstructured":"Lamaakal, I., Yahyati, C., Ouahbi, I., El Makkaoui, K., and Maleh, Y. (2025). A survey of model compression techniques for TinyML applications. 2025 International Conference on Circuit, Systems and Communication (ICCSC), IEEE.","DOI":"10.1109\/ICCSC66714.2025.11135279"},{"key":"ref_52","doi-asserted-by":"crossref","first-page":"2900","DOI":"10.1109\/TPAMI.2023.3334614","article-title":"Structured pruning for deep convolutional neural networks: A survey","volume":"46","author":"He","year":"2024","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_53","unstructured":"Hanson, S., Cowan, J., and Giles, C. (1992). Second order derivatives for network pruning: Optimal Brain Surgeon. Advances in Neural Information Processing Systems, Morgan\u2013Kaufmann. Available online: https:\/\/proceedings.neurips.cc\/paper_files\/paper\/1992\/file\/303ed4c69846ab36c2904d3ba8573050-Paper.pdf."},{"key":"ref_54","doi-asserted-by":"crossref","first-page":"9023","DOI":"10.1109\/TPAMI.2024.3416316","article-title":"Influence function based second-order channel pruning: Evaluating true loss changes for pruning is possible without retraining","volume":"46","author":"Cheng","year":"2024","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_55","doi-asserted-by":"crossref","first-page":"10558","DOI":"10.1109\/TPAMI.2024.3447085","article-title":"A survey on deep neural network pruning: Taxonomy, comparison, analysis, and recommendations","volume":"46","author":"Cheng","year":"2024","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_56","doi-asserted-by":"crossref","first-page":"801","DOI":"10.1109\/JIOT.2025.3624283","article-title":"Tiny Deep Learning Models With Hybrid Compression Techniques for Gesture-Based Air Handwriting Recognition of English Alphabets on Edge Device","volume":"13","author":"Lamaakal","year":"2026","journal-title":"IEEE Internet Things J."},{"key":"ref_57","doi-asserted-by":"crossref","first-page":"25","DOI":"10.1109\/MITP.2026.3651804","article-title":"Generative Internet of Things","volume":"28","author":"Maleh","year":"2026","journal-title":"IT Prof."}],"container-title":["Machine Learning and Knowledge Extraction"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2504-4990\/8\/5\/116\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,28]],"date-time":"2026-04-28T15:49:17Z","timestamp":1777391357000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2504-4990\/8\/5\/116"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,28]]},"references-count":57,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2026,5]]}},"alternative-id":["make8050116"],"URL":"https:\/\/doi.org\/10.3390\/make8050116","relation":{},"ISSN":["2504-4990"],"issn-type":[{"value":"2504-4990","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,28]]}}}