Home / Current Issue / Paper 1714904
Design and Performance Evaluation of a Lightweight Face Recognition Model for Android Devices
Subject area: Science,Engineering and Technology · Area of research: Mobile Computing
Abstract
Mention Face recognition systems have become increasingly important in mobile authentication, attendance monitoring, and smart surveillance applications. However, deploying deep learning–based face recognition models on mobile devices remains challenging due to limited computational resources, memory constraints, and power consumption. This research proposes a lightweight and efficient face recognition framework optimized specifically for Android-based mobile devices. The proposed system utilizes a MobileNetV2-based convolutional neural network architecture to extract discriminative facial embeddings, followed by cosine similarity for identity matching. To ensure mobile compatibility, the trained model is converted and optimized using TensorFlow Lite with post-training quantization techniques, significantly reducing model size and inference latency while maintaining high recognition accuracy. Experimental evaluation was conducted using the Labeled Faces in the Wild (LFW) dataset. The optimized lightweight model achieved competitive accuracy while reducing model size by over 70% and decreasing inference time by more than 60% compared to a standard CNN-based implementation. Real-time testing on Android devices demonstrated efficient on-device processing with low memory usage and minimal battery impact. The results indicate that lightweight deep learning architectures combined with model optimization techniques can enable accurate and real-time face recognition on resource-constrained mobile platforms. The proposed framework provides a practical and scalable solution for deploying secure biometric authentication systems on modern smartphones.
Keywords
Face Recognition, Deep Learning, MobileNetV2, TensorFlow Lite, Model Quantization, On-Device Inference, Lightweight Neural Networks, Mobile Biometrics.
References
[1] F. Schroff, D. Kalenichenko, and J. Philbin, “FaceNet: A unified embedding for face recognition and clustering,” in Proc. IEEE CVPR, 2015.
[2] M. Sandler et al., “MobileNetV2: Inverted residuals and linear bottlenecks,” in Proc. IEEE CVPR, 2018.
[3] TensorFlow Lite, “TensorFlow Lite Documentation,” Google, 2023.
[4] Y. Taigman et al., “DeepFace: Closing the gap to human-level performance in face verification,” in Proc. IEEE CVPR, 2014.
[5] J. Deng et al., “ArcFace: Additive angular margin loss for deep face recognition,” in Proc. IEEE CVPR, 2019.
[6] X. Zhang et al., “ShuffleNet: An extremely efficient convolutional neural network for mobile devices,” in Proc. IEEE CVPR, 2018.
[7] P. Viola and M. Jones, “Rapid object detection using a boosted cascade of simple features,” in Proc. IEEE CVPR, 2001.
[8] S. Han, J. Pool, J. Tran, and W. Dally, “Learning both weights and connections for efficient neural networks,” in Proc. NeurIPS, 2015.
[9] G. B. Huang, M. Ramesh, T. Berg, and E. Learned-Miller, “Labeled Faces in the Wild: A Database for Studying Face Recognition in Unconstrained Environments,” Univ. of Massachusetts, Amherst, Tech. Rep., 2007.
[10] D. P. Kingma and J. Ba, “Adam: A Method for Stochastic Optimization,” in Proc. ICLR, 2015.
[11] J. Chi, C. Kim On, H. Zhang, and S. S. Chai, “A review of deep convolutional neural networks in mobile face recognition,” Int. J. Interact. Mobile Technol., vol. 17, no. 23, pp. 4–19, Dec. 2023.
[12] A. George et al., “EdgeFace: Efficient face recognition model for edge devices,” arXiv preprint arXiv:2307.01838, 2023.
[13] A. Hassanpour and Y. Kowsari, “Lightweight face recognition: An improved MobileFaceNet model,” arXiv preprint, 2023.
How to cite this paper
@article{1714904,
author = {Mukti Gupta, Dr Choudhry Ravi Singh, Dr Gaurav Agarwal},
title = {Design and Performance Evaluation of a Lightweight Face Recognition Model for Android Devices},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {9},
pages = {629-635},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1714904.pdf},
abstract = {Mention Face recognition systems have become increasingly important in mobile authentication, attendance monitoring, and smart surveillance applications. However, deploying deep learning–based face recognition models on mobile devices remains challenging due to limited computational resources, memory constraints, and power consumption. This research proposes a lightweight and efficient face recognition framework optimized specifically for Android-based mobile devices. The proposed system utilizes a MobileNetV2-based convolutional neural network architecture to extract discriminative facial embeddings, followed by cosine similarity for identity matching. To ensure mobile compatibility, the trained model is converted and optimized using TensorFlow Lite with post-training quantization techniques, significantly reducing model size and inference latency while maintaining high recognition accuracy. Experimental evaluation was conducted using the Labeled Faces in the Wild (LFW) dataset. The optimized lightweight model achieved competitive accuracy while reducing model size by over 70% and decreasing inference time by more than 60% compared to a standard CNN-based implementation. Real-time testing on Android devices demonstrated efficient on-device processing with low memory usage and minimal battery impact. The results indicate that lightweight deep learning architectures combined with model optimization techniques can enable accurate and real-time face recognition on resource-constrained mobile platforms. The proposed framework provides a practical and scalable solution for deploying secure biometric authentication systems on modern smartphones.},
keywords = {Face Recognition, Deep Learning, MobileNetV2, TensorFlow Lite, Model Quantization, On-Device Inference, Lightweight Neural Networks, Mobile Biometrics.},
month = {March},
doi = {https://doi.org/10.64388/IREV9I9-1714904}
}