Welcome to Francis Academic Press

Academic Journal of Computing & Information Science, 2023, 6(10); doi: 10.25236/AJCIS.2023.061003.

A Face Recognition Method Using ResNet34 and RetinaFace


Lingfeng Ma, Zuqiang Long

Corresponding Author:
Lingfeng Ma

School of Physics and Electronic Engineering, Hengyang Normal University, Hengyang, Hunan, China


A new face recognition method is proposed by utilizing ResNet34 and RetinaFace, which is based on a lightweight framework for Python named Deepface. The new method is used to improve two shortcomings in the related literature: (1) the susceptibility of face recognition to interference, and (2) the quite limited number of faces detected in an image. First, the RetinaFace detector is used to replace the common detector to get more facial feature points and expand the area for detecting faces. Thus, the number of faces detected in the same image is increased. Then, ResNet34 model is applied to replace the default model in Deepface to improve the anti-interference in face recognition. Finally, experiments demonstrate that the new method is superior to the default one.


Face recognition, ResNet34, RetinaFace, Detection number of faces, Anti-interference

Cite This Paper

Lingfeng Ma, Zuqiang Long. A Face Recognition Method Using ResNet34 and RetinaFace. Academic Journal of Computing & Information Science (2023), Vol. 6, Issue 10: 18-23. https://doi.org/10.25236/AJCIS.2023.061003.


[1] Grm, Klemen, et al. "Strengths and weaknesses of deep learning models for face recognition against image degradations." Iet Biometrics 7.1 (2018): 81-89.

[2] Sun, Linjun, et al. "Gradient-enhanced softmax for face recognition." IEICE TRANSACTIONS on Information and Systems 103.5 (2020): 1185-1189.

[3] Vo, Duc My, Duc Manh Nguyen, and Sang-Woong Lee. "Deep softmax collaborative representation for robust degraded face recognition." Engineering Applications of Artificial Intelligence 97 (2021): 104052.

[4] Zhou, Shengwei, et al. "Double additive margin softmax loss for face recognition." Applied Sciences 10.1 (2019): 60.

[5] Serengil S I, Ozpinar A. Lightface: A hybrid deep face recognition framework[C]//2020 innovations in intelligent systems and applications conference (ASYU). IEEE, 2020: 1-5.

[6] Vishwakarma V P, Dalal S. A novel non-linear modifier for adaptive illumination normalization for robust face recognition [J]. Multimedia Tools and Applications, 2020, 79(17): 11503-11529.

[7] Zhu Q, He Z, Zhang T, et al. Improving classification performance of softmax loss function based on scalable batch-normalization [J]. Applied Sciences, 2020, 10(8): 2950.

[8] Wang C, Song L, Wang G, et al. Multi-scale multi-patch person re-identification with exclusivity regularized softmax[J]. Neurocomputing, 2020, 382: 64-70.

[9] Li X, Wang W. Learning discriminative features via weights-biased softmax loss [J]. Pattern Recognition, 2020, 107: 107405.

[10] Hyun J, Kim J, Choi C H, et al. Hardware architecture of a Haar classifier based face detection system using a skip scheme [C]//2021 IEEE International Symposium on Circuits and Systems (ISCAS). IEEE, 2021: 1-4.

[11] Deng X, Zhu Z, Chang J, et al. Algorithm research of face recognition system based on haar[M]//Advances in Computer Science and Ubiquitous Computing. Springer, Singapore, 2021: 317-323.

[12] Parkhi O, Vedaldi A, Zisserman A. Deep face recognition[C]//BMVC 2015-Proceedings of the British Machine Vision Conference 2015. British Machine Vision Association, 2015.

[13] Deng J, Guo J, Ververas E, et al. Retinaface: Single-shot multi-level face localisation in the wild[C] //Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 5203-5212.

[14] Deng J, Guo J, Xue N, et al. Arcface: Additive angular margin loss for deep face recognition[C] //Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 4690-4699.