Penerapan Deepface dan Retinaface dalam Pengenalan Wajah Parsial Untuk Aplikasi Keamanan Digital

Helmi Yulianti Fauziah; Tubagus Maulana Kusuma

doi:10.35760/ik.2025.v30i2.241

Authors

Helmi Yulianti Fauziah Universitas Gunadarama Author
Tubagus Maulana Kusuma Universitas Gunadarama Author

DOI:

https://doi.org/10.35760/ik.2025.v30i2.241

Keywords:

Pengenalan wajah parsial, RetinaFace, ArcFace, deep learning, biometrik

Abstract

Pengenalan wajah merupakan aspek penting dalam teknologi keamanan modern dan memiliki banyak aplikasi. Namun demikian teknologi ini juga menghadapi tantangan, terutama dalam mengidentifikasi wajah yang hanya sebagian terlihat akibat berbagai kondisi seperti tertutup masker, kacamata atau benda lain. Penelitian ini mengusulkan solusi penegnalan wajah parsial berbasis deep learning dengan mengintegrasikan RetinaFace sebagai detektor wajah dan ArcFace yang diimplementasikan melalui framework DeepFace sebagai ekstraktor fitur untuk meningkatkan akurasi identifikasi wajah tidak utuh. Dataset dikumpulkan menggunakan kamera ponsel dengan variasi pencahayaan, ekspresi, dan oklusi (masker/kacamata), kemudian diproses melalui pipeline prapengolahan citra yang mencakup konversi warna, resizing, dan normalisasi. Model dilatih dengan loss function berbasis angular margin untuk memaksimalkan jarak antar kelas, dioptimasi menggunakan Adam (learning rate 0.001) dan dievaluasi melalui confusion matrix. Hasil pengujian menunjukkan akurasi 92% pada citra statis dan 94% dalam skenario realtime, dengan kesalahan prediksi terutama terjadi pada wajah tertutup lebih dari 50%. Keunggulan sistem ini terletak pada kombinasi deteksi multi-tugas RetinaFace (landmark + bounding box) dan ekstraksi fitur diskriminatif ArcFace, yang terbukti robust terhadap variasi parsial. Temuan ini mendukung aplikasi praktis seperti absensi digital dan keamanan berbasis kamera, dengan rekomendasi peningkatan kualitas dataset dan hardware untuk mengurangi false negative.

References

Kolda, T., Bader, B., & Anderson, J. (2019). Multi-biometric system based on cutting-edge equipment for experimental contactless verification. Sensors, 19(17), 3709. https://doi.org/10.3390/s19173709

Guo, Y. (2021). Impact on biometric identification systems. Scientific Programming, 2021, 1–10. https://doi.org/10.1155/2021/3225687

Damer, N., Grebe, J. H., & Kuijper, A. (2020). The effect of wearing a mask on face recognition performance: An exploratory study. arXiv preprint, arXiv:2007.13521. https://doi.org/10.48550/arXiv.2007.13521

Huang, X., Li, L., & Wang, Y. (2023). Periocular biometric recognition for masked faces. Wuhan University Journal of Natural Sciences, 28(2), 141–152. https://doi.org/10.1051/wujns/2023282141

Choudhury, T., Dey, S., & Paul, S. (2018). A survey on biometrics and cancelable biometrics systems. International Journal of Image and Graphics, 18(4), 1850006. https://doi.org/10.1142/S0219467818500067

Habibu, M., & Sam, A. (2018). Assessment of vulnerabilities of the biometric template protection mechanism. International Journal of Advanced Technology and Engineering Exploration, 5(44), 282–288. https://doi.org/10.19101/ijatee.2018.544003

Montero, D., Chen, Z., & Rao, Y. (2021). Boosting masked face recognition with multi-task ArcFace. arXiv preprint, arXiv:2104.09874. https://doi.org/10.48550/arXiv.2104.09874

Tsai, Y.-C., Lin, H.-T., & Wang, P.-S. (2023). Joint masked face recognition and temperature measurement system using convolutional neural networks. Sensors, 23(6), 2901. https://doi.org/10.3390/s23062901

Nurlita, A., Suryani, D., & Siregar, R. (2024). Comparison of ArcFace and Dlib performance in face recognition with detection using YOLOv8. Inovtek Polbeng - Seri Informatika, 9(1), 13–21. https://doi.org/10.35314/3jy3dy73

Sydor, A., Bielecka, D., & Nowak, M. (2024). Research on the state-of-the-art deep learning based models for face detection and recognition. Information and Communications Technologies Electronic Engineering, 2, 49–55. https://doi.org/10.23939/ictee2024.02.049

Tapia, J. E., Perez, C. A., & Quezada, A. (2022). Selfie periocular verification using an efficient super-resolution approach. IEEE Access, 10, 67655–67666. https://doi.org/10.1109/ACCESS.2022.3184301

Asmitha, P., Rupa, C., & Nikitha, S. (2023). Novel and Effective Approach for Multiview Biometric Object Detection using Deep Learning based Cutting Edge Techniques. 2023 International Conference on Integrated Intelligence and Communication Systems (ICIICS), 1-6. https://doi.org/10.1109/ICIICS59993.2023.10421021.

Wang, G., Li, J., Wu, Z., Xu, J., Shen, J., & Yang, W. (2023). EfficientFace: an efficient deep network with feature enhancement for accurate face detection. Multimedia Systems, 29, 2825-2839. https://doi.org/10.1007/s00530-023-01134-6.

Jiao, J., Liu, W., Mo, Y., Jiao, J., Deng, Z., & Chen, X. (2021). Dyn-arcFace: dynamic additive angular margin loss for deep face recognition. Multimedia Tools and Applications, 80, 25741 - 25756. https://doi.org/10.1007/s11042-021-10865-5.

Deng, J., Guo, J., Liu, T., Gong, M., & Zafeiriou, S. (2020). Sub-center ArcFace: Boosting Face Recognition by Large-Scale Noisy Web Faces. , 741-757. https://doi.org/10.1007/978-3-030-58621-8_43.

R, A., Solayappan, V., T, S., & K, R. (2021). Masked Deep Face Recognition using ArcFace and Ensemble Learning. 2021 IEEE 2nd International Conference on Technology, Engineering, Management for Societal impact using Marketing, Entrepreneurship and Talent (TEMSMET), 1-6. https://doi.org/10.1109/temsmet53515.2021.9768777.

Chen, Z., Chen, J., Ding, G., & Huang, H. (2022). A lightweight CNN-based algorithm and implementation on embedded system for real-time face recognition. Multimedia Systems, 29, 129-138. https://doi.org/10.1007/s00530-022-00973-z.

Deng, J., Guo, J., & Zafeiriou, S. (2018). ArcFace: Additive Angular Margin Loss for Deep Face Recognition. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 4685-4694. https://doi.org/10.1109/CVPR.2019.00482.

Ding, F., Peng, P., Huang, Y., Geng, M., & Tian, Y. (2020). Masked Face Recognition with Latent Part Detection. Proceedings of the 28th ACM International Conference on Multimedia. https://doi.org/10.1145/3394171.3413731.

An, X., Zhu, X., Xiao, Y., Wu, L., Zhang, M., Gao, Y., Qin, B., Zhang, D., & Fu, Y. (2020). Partial FC: Training 10 Million Identities on a Single Machine. 2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), 1445-1449. https://doi.org/10.1109/ICCVW54120.2021.00166.