The article considers the problem of determining the state of human fatigue based on the analysis of graphical images of pupillograms using convolutional neural networks. The relevance of the study is determined by the need to develop automated systems for monitoring the functional state of a person, since the state of fatigue can lead to a decrease in concentration, a deterioration in performance and an increase in the likelihood of dangerous situations. The initial data for the analysis was our own experimental dataset of pupillogram images, consisting of 831 images, divided into two classes: Normal (369 images) and Deviation (462 images). All images were obtained using a software package for determining the functional state of a person based on the pupillary reaction to light exposure. Examples of pupillogram images corresponding to each of the classes are presented. The paper provides a comparative analysis of various architectures of pre-trained convolutional neural networks: ResNet50, InceptionV3, EfficientNetV2 and ConvNeXt. In addition, a one-dimensional convolutional neural network (1D CNN) was used to solve the problem. It was compared with pre-trained models. All of the models were trained under identical conditions using the same hyperparameters and data preprocessing methods, ensuring the accuracy and comparison of results. Model quality was assessed using the Accuracy and F1-score metrics, as well as an analysis of the classification error matrix. Particular attention was paid to False Negative errors, as missing fatigue is critical in human condition monitoring tasks. The study found that EfficientNetV2 was the most effective model, demonstrating the best classification performance and the absence of False Negative errors. It was also found that modern convolutional neural network architectures provide higher efficiency compared to earlier models. Using a one-dimensional convolutional neural network (1D CNN) with image transformation into one-dimensional sequences yielded lower results, confirming the importance of preserving the spatial structure of pupillogram images when solving the classification problem. Overall, the obtained results of the study confirmed the feasibility of using modern convolutional neural networks for solving the problem of determining human fatigue based on pupillogram image analysis.
CONVOLUTIONAL NEURAL NETWORK, CONVOLUTIONAL NEURAL NETWORK MODEL, PRETRAINED MODEL, PUPILLOGRAM, IMAGE CLASSIFICATION, FATIGUE DETECTION
1. A.O. Bulygin, A.M. Kashevnik, Data Analysis and Processing Systems, 3 (83), 19-36 (2021).
2. A.A. Sibgatullin, A.S. Katasev, Bulletin of the Technological University, 25, 12, 139-143 (2022).
3. O.L. Isaeva, M.P. Boronenko, Modern science-intensive technologies, 3, 54-58 (2020).
4. O.L. Isaeva, M.P. Boronenko, Bulletin of the Yugra State University, 3, 59-67 (2021).
5. B.V. Kostrov, S.N. Baranova, A.A. Vyugina, Y.A. Trushina, I.S. Panina, International Research Journal, 8, 1-7 (2024).
6. R.N. Minnikhanov, T.R. Batorshin, R.M. Gabbazov, R.I. Fakhraziev, A.S. Katasev, Bulletin of Voronezh State University. Series: Systems Analysis and Information Technology, 3, 110-122 (2025).
7. P.I. Vladimirov, E.Yu. Zykov, V.V. Kugurakova, Electronic libraries, 23, 6, 1121-1141 (2020).
8. R.N. Minnikhanov, A.S. Katasev, D.V. Kataseva, V. Skibin, Bulletin of the Technological University, 27, 11, 199-205 (2024).
9. D.A. Nabiullin, V.V. Kononova, S.V. Novikova, Bulletin of the Technological University, 24, 6, 103-107 (2021).
10. K.A. Matveeva, R.N. Minnikhanov, A.S. Katasev, Bulletin of the Technological University, 27, 1, 76-80 (2024).
11. A.A. Voevoda, V.I. Shipagin, Data Analysis and Processing Systems, 4 (88), 7-30 (2022).
12. D.V. Kataseva, D.D. Loseva, Bulletin of the Technological University, 25, 1, 77-80 (2022).
13. T.Yu. Gainutdinova, S.V. Novikova, V.G. Gainutdinov, N.V. Levshonkov, News of higher educational institutions. Aviation technology, 4, 54-61 (2023).
14. D.V. Kataseva, E.A. Bakaldina, Bulletin of the Technological University, 29, 1, 113-119 (2026).
15. O.A. Pyrnova, Scientific and Technical Bulletin of the Volga Region, 11, 254-256 (2023).
16. S.A. Yarushev, A.N. Lukyanov, D.A. Vorobyov, A.A. Polyakov, Soft Measurements and Computing, 86, 1, 34-41 (2025).
17. I.N. Barskaya, K. Imanbekov, G.L. Kushnir, Original research, 15, 4, 190-195 (2025).
18. D.G. Petrosyants, A.S. Katasev, L.Yu. Emaletdinova, Scientific and Technical Bulletin of the Volga Region, 6, 85-87 (2023).
19. K.E. Chertilin, V.D. Ivchenko, Devices, 7 (265), 37-47 (2022).
20. D.P. Nikonorov, A.S. Katasev, Bulletin of the Technological University, 28, 6, 110-114 (2025).
21. M.T. Roseno, S. Oktarina, Yu. Nearti, H. Syaputra, N. Jayanti, Journal of Information Systems and Informatics, 6, 3, 2099-2109 (2024).
22. S. Gupta, A.B.A. Hamid, T.E. Nyamasvisva, V. Jain, N. Tyagi, Ng.Kh. Mun, D. Ather, International Journal of Engineering, Science and Information Technology, 5, 3, 588-594 (2025).
23. R.K. Futri, M. Tarigan, Jurnal Mahasiswa Teknik Informatika, 9, 3, 4135-4143 (2025).
24. Le. Zhang, W. Song, T. Zhu, Ya. Liu, W. Chen, Ya. Cao, Briefings in Bioinformatics, 25, 3 (2024).
25. Ch. Kondaiah, A.R. Pais, R.S. Rao, IEEE Transactions on Information Forensics and Security, 20, 12188-12198 (2025).
26. D.V. Kataseva, A.A. Alekseeva, Bulletin of the Technological University, 29, 3, 130-135 (2026).
27. F.Yu. Kopylov, O.A. Pyrnova, Soft measurements and calculations, 98, 1-2, 5-14 (2026).
28. V.A. Arkhipov, Bulletin of the Altai Academy of Economics and Law, 9-2, 12-15 (2019).
29. A.S. Katasev, D.V. Kataseva, Yu.N. Smirnov, A.A. Sibgatullin, Bulletin of the Technological University, 27, 7, 126-130 (2024).
30. V.A. Bulanov, O.E. Fomicheva, Devices and systems. Management, control, diagnostics, 2, 40-47 (2020).
31. D.V. Gordienko, A.O. Kravchenko, Digital Diagnostics, 5, S1, 109-111 (2024).
32. A.N. Matveev, International Conference on Soft Computing and Measurements, 1, 166-168 (2023).
33. A.N. Aparnev, O.V. Bartenyev, Bulletin of the Moscow Power Engineering Institute, 2, 90-105 (2020).
34. Y. Zhao, K. Xie, Z. Zou, J.-B. He, IEEE Access, 8, 144205-144217 (2020).
35. F.M. Shiri, E. Ahmadi, M. Rezaee, T. Perumal, Journal on Artificial Intelligence, 6, 1, 85-103 (2024).
36. M.E. Kolobaev, D.A. Prytyka, D.O. Mikhailova, E.I. Voevodina, A.V. Yurchenko, Soft Measurements and Computing, 69, 8, 69-78 (2023).
37. A.S. Prokopenya, I.S. Azarov, Big Data and Advanced Analytics, 6-1, 271-280 (2020).
38. N.R. Kashapov, A.S. Katasev, D.V. Kataseva, Information and Security, 19, 4, 555-558 (2016).
39. A.S. Katasev, T.I. Tukhbatullin, Bulletin of the Technological University, 26, 4, 53-57 (2023).
40. V.V. Erokhin, Soft Measurements and Computing, 48, 11, 66-83 (2021).
