10.14489/vkit.2025.10.pp.010-021

DOI: 10.14489/vkit.2025.10.pp.010-021

Разуваев А. В., Осанов В. А.
АНАЛИЗ МЕТОДОВ ДЕТЕКЦИИ ДЛЯ РАСПОЗНАВАНИЯ ЛИЧНОСТИ НА ИЗОБРАЖЕНИИ
(pp.10-21)

Аннотация. Рассмотрены современные методы детекции лиц, основанные на различных алгоритмах и архитектурах нейронных сетей. Описаны популярные детекторы, включая классический метод каскадных классификаторов Хаара, реализованный в OpenCV, а также современные нейросетевые подходы: SSD, MTCNN, RetinaFace, YUNet и Dlib. Рассмотрены их особенности, принципы работы и области применения. Особое внимание уделено эффективности и точности обнаружения лиц в сложных условиях, таких как цветовые искажения, изменяющееся освещение, окклюзия. Представленные методы сравнивались по скорости работы, вычислительным требованиям и качеству детекции. Проведен анализ преимуществ и недостатков различных подходов, что позволяет определить наиболее подходящий метод в зависимости от требований к точности, скорости и вычислительным ресурсам.

Ключевые слова: детекция лиц; детекторы; нейронные сети; каскадные классификаторы Хаара; OpenCV; SSD; MTCNN; RetinaFace; YUNet; Dlib.

Razuvaev A. V., Osanov V. A.
ANALYSIS OF DETECTION METHODS FOR PERSONAL RECOGNITION IN IMAGES
(pp.10-21)

Abstract. This article examines modern methods of face detection based on various algorithms and architectures of neural networks. Popular detectors are presented, including the classic cascade Haar classifier method implemented in OpenCV, as well as modern neural network approaches: SSD, MTCNN, RetinaFace, YUNet and Dlib. Their features, principles of operation and areas of application are considered, which makes it possible to evaluate their effectiveness in various use scenarios. Special attention is paid to the accuracy and reliability of face detection in difficult conditions, such as various color distortions, lighting changes, partial occlusion, as well as the presence of noise or foreign objects in the frame. The article presents a comparative analysis of the considered methods by key parameters.: speed of operation, computational requirements, and detection quality. The cascade Haar classifier, despite its simplicity and high speed of operation, is inferior in accuracy to modern neural network methods, especially under difficult shooting conditions. SSD and MTCNN provide a balance between performance and detection quality, making them suitable for mobile devices and real-time systems. RetinaFace, in turn, demonstrates outstanding accuracy even under difficult conditions, but requires significant computing resources. Dlib and YUNet offer compromise solutions suitable for various tasks, from fast image processing to high-precision detection. The possibilities of adapting and optimizing algorithms for specific hardware platforms such as mobile devices, embedded systems, and server solutions are also being considered. The article provides recommendations on choosing the most appropriate face detection method, depending on the requirements for accuracy, speed, and available computing resources, which makes it useful for computer vision developers and machine learning specialists.

Keywords: Face detection; Detectors; Neural networks; Cascading Haar classifiers; OpenCV; SSD; MTCNN; RetinaFace; YUNet; Dlib.

+ - Информация об авторах (About the Authors) Click to collapse

Рус

А. В. Разуваев, В. А. Осанов (Поволжский государственный университет телекоммуникаций и информатики, Самара, Россия) E-mail: Этот e-mail адрес защищен от спам-ботов, для его просмотра у Вас должен быть включен Javascript

Eng

A. V. Razuvaev, V. A. Osanov (Povolzhskiy State University of Telecommunications and Informatics, Samara, Russia) E-mail: Этот e-mail адрес защищен от спам-ботов, для его просмотра у Вас должен быть включен Javascript

+ - Библиографический список (References) Click to collapse

Рус

1. Prasantha S. Detailed survey of machine learning algorithms for face recognition // International journal of creative research thoughts. 2023. V. 11, No. 11. P. 832–836.
2. Liu R. Face recognition based on Convolutional Neural Networks // Highlights in Science, Engineering and Technology. 2022. V. 16. P. 32–39. DOI: 10.54097/hset.v16i.2225
3. Dergachov K., Krasnov L., Cheliadin O., Zymovin A. Adaptive algorithms of face detection and effectiveness assessment of their use // Advanced Information Systems. 2018. V. 2, No. 3. P. 10–18. DOI: 10.20998/2522-9052.2018.3.02
4. Advancing cybersecurity: a comprehensive review of AI-driven detection techniques / A. H. Salem, S. M. Azzam, O. E. Emam et al. // Journal of Big Data. 2024. V. 11, No. 105. P. 1–38. DOI: 10.1186/s40537-024-00957-y
5. Musikhin A., Burenin S. Face recognition using multitasking cascading convolutional networks // IOP Conference Series: Materials Science and Engineering. 29–30 April 2021. Krasnoyarsk, Russia. Р. 1–5. DOI: 10.1088/1757-899X/1155/1/012057
6. Kaur H., Mirza A. Face Detection Using Haar Cascades Classifier // 2nd International Conference on ICT for Digital, Smart, and Sustainable Development (ICIDSSD). 27–28 February 2021. New Delhi, India. P. 1–7. DOI: 10.4108/eai.27-2-2020.2303218
7. Wu W., Peng H., Yu S. YuNet: A Tiny Millisecond-Level Face Detector // Machine Intelligence Research. 2023. V. 20. P. 656–665. DOI: 10.1007/s11633-023-1423-y
8. Aydın M. T., Menemencioğlu O., Orak İ. M. Face recognition approach using DLIB and K-NN // Current trends in computing. 2023. V. 1, No. 2. P. 93–103.

Eng

1. Prasantha, S. (2023). Detailed survey of machine learning algorithms for face recognition. International Journal of Creative Research Thoughts, 11(11), 832–836.
2. Liu, R. (2022). Face recognition based on Convolutional Neural Networks. Highlights in Science, Engineering and Technology, 16, 32–39. https://doi.org/10.54097/hset.v16i.2225
3. Dergachov, K., Krasnov, L., Cheliadin, O., & Zymovin, A. (2018). Adaptive algorithms of face detection and effectiveness assessment of their use. Advanced Information Systems, 2(3), 10–18. [in Russian language] https://doi.org/10.20998/2522-9052.2018.3.02
4. Salem, A. H., Azzam, S. M., Emam, O. E., et al. (2024). Advancing cybersecurity: A comprehensive review of AI-driven detection techniques. Journal of Big Data, 11(105), 1–38. https://doi.org/10.1186/s40537-024-00957-y
5. Musikhin, A., & Burenin, S. (2021). Face recognition using multitasking cascading convolutional networks. IOP Conference Series: Materials Science and Engineering, 1155(1), 1–5. [in Russian language] https://doi.org/10.1088/1757-899X/1155/1/012057
6. Kaur, H., & Mirza, A. (2021, February 27–28). Face detection using Haar Cascades classifier [Paper presentation]. 2nd International Conference on ICT for Digital, Smart, and Sustainable Development (ICIDSSD), New Delhi, India. https://doi.org/10.4108/eai.27-2-2020.2303218
7. Wu, W., Peng, H., & Yu, S. (2023). YuNet: A tiny millisecond-level face detector. Machine Intelligence Research, 20, 656–665. https://doi.org/10.1007/s11633-023-1423-y
8. Aydın, M. T., Menemencioğlu, O., & Orak, İ. M. (2023). Face recognition approach using DLIB and K-NN. Current Trends in Computing, 1(2), 93–103.

+ - Заказать электронную версию статьи (Purchase digital version of a single article) Click to collapse

Рус

Статью можно приобрести в электронном виде (PDF формат).

Стоимость статьи 700 руб. (в том числе НДС 20%). После оформления заказа, в течение нескольких дней, на указанный вами e-mail придут счет и квитанция для оплаты в банке.

После поступления денег на счет издательства, вам будет выслан электронный вариант статьи.

Для заказа скопируйте doi статьи:

10.14489/vkit.2025.10.pp.010-021

и заполните форму

Отправляя форму вы даете согласие на обработку персональных данных.

Eng

This article is available in electronic format (PDF).

The cost of a single article is 700 rubles. (including VAT 20%). After you place an order within a few days, you will receive following documents to your specified e-mail: account on payment and receipt to pay in the bank.

After depositing your payment on our bank account we send you file of the article by e-mail.

To order articles please copy the article doi:

10.14489/vkit.2025.10.pp.010-021

and fill out the form