Portrait age recognition method based on improved ResNet and deformable convolution

Ji Xi; Zhe Xu; Zihan Yan; Wenjie Liu; Yanting Liu; Ji Xi; Zhe Xu; Zihan Yan; Wenjie Liu; Yanting Liu

doi:10.3934/era.2023333

Electronic Research Archive

2023, Volume 31, Issue 11: 6585-6599. doi: 10.3934/era.2023333

Previous Article Next Article

Research article Special Issues

Portrait age recognition method based on improved ResNet and deformable convolution

1.
School of Computer Information Engineering, Changzhou Institute of Technology, Changzhou 213022, China
2.
School of Software and Big Data, Changzhou College of Information Technology, Changzhou 213032, China

Received: 29 June 2023 Revised: 26 September 2023 Accepted: 27 September 2023 Published: 13 October 2023

ResNet-based correlation models excel in age recognition algorithms, but specific age recognition research is currently limited and often plagued by substantial errors. We introduce an enhanced portrait age recognition algorithm based on ResNet, using CORAL (consistent rank logits) rank consistent ordered regression instead of traditional classification to predict precise ages. We further improve this approach by incorporating DCN (deformable convolution), resulting in the DCN-R model. DCN dynamically adjusts convolution kernels for diverse faces, improving accuracy and robustness. We tested DCN-R34 and DCN-R50 against the SOTA model, achieving better results with the same complexity. This reduces the computational load while maintaining or enhancing performance.
- ResNet,
- deformable convolution,
- portrait age recognition,
- rank-uniform ordered regression
Citation: Ji Xi, Zhe Xu, Zihan Yan, Wenjie Liu, Yanting Liu. Portrait age recognition method based on improved ResNet and deformable convolution[J]. Electronic Research Archive, 2023, 31(11): 6585-6599. doi: 10.3934/era.2023333

Related Papers:

Abstract

ResNet-based correlation models excel in age recognition algorithms, but specific age recognition research is currently limited and often plagued by substantial errors. We introduce an enhanced portrait age recognition algorithm based on ResNet, using CORAL (consistent rank logits) rank consistent ordered regression instead of traditional classification to predict precise ages. We further improve this approach by incorporating DCN (deformable convolution), resulting in the DCN-R model. DCN dynamically adjusts convolution kernels for diverse faces, improving accuracy and robustness. We tested DCN-R34 and DCN-R50 against the SOTA model, achieving better results with the same complexity. This reduces the computational load while maintaining or enhancing performance.

References

[1]	Z. Huang, J. Zhang, H. Shan, When age-invariant face recognition meets face age synthesis: A multi-task learning framework and a new benchmark, IEEE Trans. Pattern Anal. Mach. Intell., 45 (2023), 7917–7932. https://doi.org/10.1109/TPAMI.2022.3217882 doi: 10.1109/TPAMI.2022.3217882
[2]	A. M. Abu Nada, E. Alajrami, A. A. Al-Saqqa, S. Abu-Naser, Age and gender prediction and validation through single user images using CNN, Int. J. Acad. Eng. Res., 4 (2020), 21–24.
[3]	I. Rafique, A. Hamid, S. Naseer, M. Asad, M. Awais, T. Yasir, Age and gender prediction using deep convolutional neural networks, in 2019 International Conference on Innovative Computing (ICIC), 2019, 1–6. https://doi.org/10.1109/ICIC48496.2019.8966704.
[4]	A. Othmani, A. R. Taleb, H. Abdelkawy, A. Hadid, Age estimation from faces using deep learning: A comparative analysis, Comput. Vision Image Understanding, 196 (2020). https://doi.org/10.1016/j.cviu.2020.102961 doi: 10.1016/j.cviu.2020.102961
[5]	N. Sharma, R. Sharma, N. Jindal, Face-based age and gender estimation using improved convolutional neural network approach, Wireless Pers. Commun., 124 (2022), 3035–3054. https://doi.org/10.1007/s11277-022-09501-8 doi: 10.1007/s11277-022-09501-8
[6]	A. Sakata, N. Takemura, Y. Yagi, Gait-based age estimation using multi-stage convolutional neural network, IPSJ Trans. Comput. Vision Appl., 4 (2019), 1–10. https://doi.org/10.1186/s41074-019-0054-2 doi: 10.1186/s41074-019-0054-2
[7]	C. Y. Hsu, L. E. Lin, C. H. Lin, Age and gender recognition with random occluded data augmentation on facial images, Multimedia Tools Appl., 80 (2021), 11631–11653. https://doi.org/10.1007/s11042-020-10141-y doi: 10.1007/s11042-020-10141-y
[8]	B. B. Mamatkulovich, H. A. Alijon o'g'li, Facial image-based gender and age estimation, Eurasian Sci. Her., 18 (2023), 47–50.
[9]	L. Li, H. T. Lin, Ordinal regression by extended binary classification, Adv. Neural Inf. Process. Syst., 19 (2006).
[10]	W. Cao, V. Mirjalili, S. Raschka, Rank consistent ordinal regression for neural networks with application to age estimation, Pattern Recognit. Lett., 140 (2020), 325–331. https://doi.org/10.1016/j.patrec.2020.11.008 doi: 10.1016/j.patrec.2020.11.008
[11]	J. Paplham, V. Franc, Unraveling the age estimation puzzle: Comparative analysis of deep learning approaches for facial age estimation, arXiv preprint, (2023), arXiv: 2307.04570. https://doi.org/10.48550/arXiv.2307.04570
[12]	J. Dai, H. Qi, Y. Xiong, Y. Li, G. Zhang, H. Hu, et al., Deformable convolutional networks, in Proceedings of the IEEE International Conference on Computer Vision, (2017), 764–773.
[13]	Z. Niu, M. Zhou, L. Wang, X. Gao, G. Hua, Ordinal regression with multiple output CNN for age estimation, in Proceedings of the IEEE conference on computer Vision and Pattern Recognition, (2016), 4920–4928.

Reader Comments

Your name:*

Email:*
© 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)