Face Detection and Identification Using Convolutional Neural Network and MobileNetV3 Model
DOI:
https://doi.org/10.31861/sisiot2025.2.02009Keywords:
machine learning, computer vision, object detection, convolutional neural networks, transfer learningAbstract
This paper presents the results of a study of the effectiveness of applying the transfer learning methodology to the task of face detection and recognition, with a focus on processing images containing only one face. MobileNetV3 was chosen as the basic neural network architecture, which provides high performance with limited computing resources. The model was trained in two consecutive stages: the first is face recognition (detection) in photos, and the second is face identification from a face image. To ensure the unambiguity and purity of the training data, only images with one face were used. The training process combined the use of an open dataset for the initial stage of face detection with the proprietary photo set of students of Yuriy Fedkovych Chernivtsi National University for the identification phase. The model was trained and tested in the Google Colab cloud environment using an Nvidia Tesla T4 GPU. The neural network was implemented using the modern deep learning framework TensorFlow and our own program code written in Python. The model parameters were optimized by minimizing the loss function, which is the sum of the binary cross-entropy and the negative logarithm of the Intersection over Union metric, which characterizes the accuracy of determining the location of an object in an image. The built model was compared with previous approaches to face detection implemented on the basis of the OpenCV library. A comparative analysis by the metrics of recognition accuracy and processing time demonstrated the superiority of the developed system. The results obtained are of interest to researchers in the field of computer vision, automated recognition systems, and technologies for intelligent visual data processing.
Downloads
References
A. Howard, M. Sandler, G. Chu, and L.-C. Chen, “Searching for MobileNetV3,” in Proc. IEEE/CVF Int. Conf. Comput. Vis. (ICCV), 2019.
M. Abadi, A. Agarwal, P. Barham, and E. Brevdo, “TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems,” arXiv preprint. arXiv:1603.04467, 2016.
O. Russakovsky, J. Deng, H. Su, and J. Krause, “ImageNet Large Scale Visual Recognition Challenge,” Int. J. Comput. Vis., vol. 115, no. 3, pp. 211–252, 2015.
M. Watson, D. Shivakumar, F. Chollet, and M. Gorner, “KerasCV and KerasNLP: Vision and Language Power-Ups,” J. Mach. Learn. Res., vol. 25, no. 375, pp. 1–10, 2024.
J. Hu, L. Shen, S. Albanie, and G. Sun, “Squeeze-and-Excitation Networks,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 42, no. 8, pp. 2011–2023, 2020.
S. Pydimarry, S. Khairnar, S. Palacios, and G. Sankaranarayanan, “Evaluating Model Performance with Hard-Swish Activation Function Adjustments,” arXiv preprint. arXiv:2410.06879, 2024.
J. He, S. Erfani, X. Ma, and J. Bailey, “Alpha-IoU: A Family of Power Intersection over Union Losses for Bounding Box Regression,” arXiv preprint. arXiv:2110.13675, 2021.
D. Kingma and J. Ba, “Adam: A Method for Stochastic Optimization,” arXiv preprint. arXiv:1412.6980, 2014.
F. Elmenshawii, “Face Detection Dataset,” [Online]. Available: https://www.kaggle.com/datasets/fareselmenshawii/face-detection-dataset.
M. Ilashchuk, I. Kushnir, and S. Melnychuk, “Rozpiznavannia oblych v realnomu chasi za dopomohoiu biblioteky OpenCV ta movy prohramuvannia Python,” Herald of Khmelnytskyi Natl. Univ., no. 341, pp. 5–21, 2024 (in Ukrainian).
Published
Issue
Section
License
Copyright (c) 2025 Security of Infocommunication Systems and Internet of Things
This work is licensed under a Creative Commons Attribution 4.0 International License.
