基于标记适应的人脸年龄识别优化算法

doi:10.19678/j.issn.1000-3428.0068561

摘要/Abstract

摘要：

缺乏完整和足够的人脸年龄标记数据集是当前人脸年龄识别问题中最突出的挑战之一。由于相近年龄的面部具有相似性, 因此在年龄识别中可以学习并利用相邻年龄的面部信息, 将每张人脸图像看作相关年龄的标记分布(LD), 从而有效缓解了训练和测试数据集不足的问题。但是在不同年龄阶段, 人的面部衰老变化速度显著不同, 如儿童和老年时期面部变化较快, 而中年时期面部变化较平缓, 当前常用的LD方法存在模式单一的缺点, 难以适应不同年龄阶段人脸特征的变化规律。为提高人脸年龄识别算法的通用性, 提出一种深度学习框架下基于标记适应的人脸年龄识别优化算法IFAR-LA。引入标记适应机制, 能够更好地学习特征表示, 挖掘人脸图像数据中丰富的语义信息, 从而有效提取不同年龄阶段面部变化的特征, 大幅提升表示学习能力和泛化能力。改进后的标记适应算法能够适应不同年龄阶段人脸变化规律, 使每幅人脸图像在学习其真实年龄和相关年龄的时候都能发挥作用, 缓解训练数据不足的问题, 同时, 提升了算法普适性, 能够适应不同年龄阶段人的面部衰老变化速度。在公开的人脸数据集MORPH和FG-NET上的实验结果表明, IFAR-LA算法相比改进前的人脸年龄识别算法平均绝对误差分别降低了6.5%和11.5%。

关键词: 标记分布学习, 深度学习, 卷积神经网络, 人脸年龄识别, 平均绝对误差

Abstract:

The lack of a complete and sufficient dataset with facial age labels is one of the most prominent challenges in current facial age recognition. Owing to the similarity between faces of similar ages, in age recognition, the facial information of neighboring ages can be utilized to treat each face image as a Label Distribution (LD) of relevant ages, thus effectively alleviating the problem of insufficient training and testing datasets. However, the aging changes in individuals' faces differ significantly at different age stages; for example, facial changes occur more rapidly in children and old age, whereas they occur more gradually during middle age. The current commonly used LD method has the disadvantage of a single mode, which makes it difficult to adapt to the change law of facial features at different age stages. To enhance the generality of the facial age recognition algorithm, an improved algorithm for facial age recognition based on label adaptation within the framework of deep learning IFAR-LA is proposed. The algorithm introduces a label adaptation mechanism, which can better learn the feature representation and mine the rich semantic information in the face image data to adapt to the features of facial changes at different ages and significantly improve the representation learning and generalization abilities. The improved label adaptation algorithm is able to adapt to the age-related changes in different stages of people's faces, so that every facial image can contribute to learning the true age and related ages of the person, greatly alleviating the problem of insufficient training data. At the same time, the algorithm has improved its universality and can adapt to the changing speed of facial aging of people of different ages. The experimental results on publicly available facial datasets MORPH and FG-NET indicate that the Mean Absolute Error(MAE) of the IFAR-LA algorithm is reduced by 6.5% and 11.5%, respectively, compared to before the improvement.

Key words: Label Distribution(LD) learning, deep learning, Convolutional Neural Network(CNN), facial age recognition, Mean Absolute Error(MAE)

张会影, 圣文顺. 基于标记适应的人脸年龄识别优化算法[J]. 计算机工程, 2025, 51(1): 174-181.

ZHANG Huiying, SHENG Wenshun. Improved Algorithm for Facial Age Recognition Based on Label Adaptation[J]. Computer Engineering, 2025, 51(1): 174-181.

https://www.ecice06.com/CN/Y2025/V51/I1/174

图/表 8

图1 不同年龄阶段人的面部变化

Fig.1 Facial appearance changes of people at different age stages

图2 基于标记适应的人脸年龄识别框架

Fig.2 Facial age recognition framework based on label adaptation

图3 人脸图像预处理

Fig.3 Facial image preprocessing

图4 在MORPH数据集上CS曲线对比

Fig.4 Comparison of CS curves on MORPH dataset

图5 在FG-NET数据集上CS曲线对比

Fig.5 Comparison of CS curves on FG-NET dataset

参考文献 30

1	ZHANG B C , BAO Y . Cross-dataset learning for age estimation. IEEE Access, 2022, 10, 24048- 24055. doi: 10.1109/ACCESS.2022.3154403
2	SUN H M , PAN H Y , HAN H , et al. Deep conditional distribution learning for age estimation. IEEE Transactions on Information Forensics and Security, 2021, 16, 4679- 4690. doi: 10.1109/TIFS.2021.3114066
3	CUI X , ZHANG Z F , CAO J , et al. Combined deep learning with directed acyclic graph SVM for local adjustment of age estimation. IEEE Access, 2021, 9, 370- 379. doi: 10.1109/ACCESS.2020.3046661
4	LIU H , LU J , FENG J J , et al. Ordinal deep learning for facial age estimation. IEEE Transactions on Circuits[WT《Times New Roman》]& Systems for Video Technology, 2019, 29 (2): 486- 501.
5	GENG X , ZHOU Z H , SMITH-MILES K . Automatic age estimation based on facial aging patterns. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29 (12): 2234- 2240. doi: 10.1109/TPAMI.2007.70733
6	ZAGHBANI S , BOUJNEH N , BOUHLEL M S . Age estimation using deep learning. Computers[WT《Times New Roman》]& Electrical Engineering, 2018, 68, 337- 347.
7	DAGHER I , BARBARA D . Facial age estimation using pre-trained CNN and transfer learning. Multimedia Tools and Applications, 2021, 80 (13): 20369- 20380. doi: 10.1007/s11042-021-10739-w
8	OTHMANI A , TALEB A R , ABDELKAWY H , et al. Age estimation from faces using deep learning: acomparative analysis. Computer Vision and Image Understanding, 2020, 196, 102961- 102961. doi: 10.1016/j.cviu.2020.102961
9	AHMED M, VIRIRI S. Deep learning using Bayesian optimization for facial age estimation[C]//Proceedings of International Conference on Image Analysis and Recognition. Berlin, Germany: Springer, 2019: 243-254.
10	GENG X , YIN C , ZHOU Z H . Facial age estimation by learning from label distributions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35 (10): 2401- 2412. doi: 10.1109/TPAMI.2013.51
11	ZHANG H Y , ZHANG Y , GENG X . Practical age estimation using deep label distribution learning. Frontiers of Computer Science, 2020, 15 (3): 153318.
12	WINDHAGER S , MITTEROECKER P , RUPIĆI , et al. Facial aging trajectories: acommon shape pattern in male and female faces is disrupted after menopause. American Journal of Physical Anthropology, 2019, 169 (4): 678- 688. doi: 10.1002/ajpa.23878
13	ANIL J , SURESH L P . A novel fast hybrid face recognition approach using convolutional Kernel extreme learning machine with HOG feature extractor. Measurement: Sensors, 2023, 30, 100907. doi: 10.1016/j.measen.2023.100907
14	PONTES J K , BRITTO A S , FOOKES C , et al. A flexible hierarchical approach for facial age estimation based on multiple features. Pattern Recognition, 2016, 54, 34- 51. doi: 10.1016/j.patcog.2015.12.003
15	SPIZHEVOI A S , BOVYRIN A V . Estimating human age using bio-inspired features and the ranking method. Pattern Recognition and Image Analysis, 2015, 25 (3): 547- 552. doi: 10.1134/S1054661815030244
16	AMNA RIZWAN S , YASIN GHADI Y , JALAL A , et al. Automated facial expression recognition and age estimation using deep learning. Computers, Materials[WT《Times New Roman》]& Continua, 2022, 71 (3): 5235- 5252.
17	YI D, LEI Z, LI S Z. Age estimation by multi-scale convolutional network[C]//Proceedings of Asian Conference on Computer Vision. Berlin, Germany: Springer, 2015: 144-158.
18	ROTHE R , TIMOFTE R , VAN GOOL L . Deep expectation of real and apparent age from a single image without facial landmarks. International Journal of Computer Vision, 2018, 126 (2): 144- 157.
19	ZHANG H Y , GENG X , ZHANG Y . Recurrent age estimation. Pattern Recognition Letters, 2019, 125, 271- 277. doi: 10.1016/j.patrec.2019.05.002
20	GENG X, WANG Q, XIA Y. Facial age estimation by adaptive label distribution learning[C]//Proceedings of the 22nd International Conference on Pattern Recognition. Washington D.C., USA: IEEE Press, 2014: 4465-4470.
21	GAO B B , XING C , XIE C W , et al. Deep label distribution learning with label ambiguity. IEEE Transactions on Image Processing, 2017, 26 (6): 2825- 2838. doi: 10.1109/TIP.2017.2689998
22	吕晓琪, 李浩, 谷宇. 基于深度学习算法的人脸图像活体特征变换尺度提取. 吉林大学学报(工学版), 2023, 53 (11): 3201- 3206.
	LÜ X Q , LI H , GU Y . Adaptive blur and deduplication algorithm for digital media image based on wavelet domain. Journal of Jilin University (Engineering and Technology Edition), 2023, 53 (11): 3201- 3206.
23	SINGH A , KANT C . Face and age recognition using three-dimensional discrete wavelet transform and rotational local binary pattern with radial basis function support vector machine method. International Journal of Electrical Engineering[WT《Times New Roman》]& Education, 2023, 60 (2): 389- 404.
24	刘成, 曹良才, 靳业, 等. 基于Transformer的跨年龄人脸识别方法. 激光与光电子学进展, 2023, 60 (10): 1010019.
	LIU C , CAO L C , JIN Y , et al. Transformer for age-invariant face recognition. Laser[WT《Times New Roman》]& Optoelectronics Progress, 2023, 60 (10): 1010019.
25	RICANEK K, TESAFAYE T. MORPH: a longitudinal image database of normal adult age-progression[C]//Proceedings of the 7th International Conference on Automatic Face and Gesture Recognition. Washington D.C., USA: IEEE Press, 2006, 341-345.
26	LANITIS A , TAYLOR C J , COOTES T F . Toward automatic simulation of aging effects on face images. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 24 (4): 442- 455. doi: 10.1109/34.993553
27	SZEGEDY C, IOFFE S, VANHOUCKE V, et al. Inception-v4, Inception-ResNet and the impact of residual connections on learning[C]//Proceedings of the AAAI Conference on Artificial Intelligence. [S. l. ]: AAAI Press, 2017: 67-71.
28	LU J W , LIONG V E , ZHOU J . Cost-sensitive local binary feature learning for facial age estimation. IEEE Transactions on Image Processing, 2015, 24 (12): 5356- 5368. doi: 10.1109/TIP.2015.2481327
29	LIU H, LU J W, FENG J J, et al. Ordinal deep feature learning for facial age estimation[C]//Proceedings of the 12th IEEE International Conference on Automatic Face[WT《Times New Roman》]& Gesture Recognition (FG 2017). Washington D. C., USA: IEEE Press, 2017: 157-164.
30	LIU X H , ZOU Y , KUANG H L , et al. Face image age estimation based on data augmentation and lightweight convolutional neural network. Symmetry, 2020, 12 (1): 146. doi: 10.3390/sym12010146

[1]	周宇, 谢威, 邝得互, 江健民. 基于三元自注意力的视频快照压缩成像重建[J]. 计算机工程, 2025, 51(1): 20-30.
[2]	胡升龙, 陈彬, 张开华, 宋慧慧. 场景结构知识增强的协同显著性目标检测[J]. 计算机工程, 2025, 51(1): 31-41.
[3]	喻勇涛, 孙奥, 李昂, 朱琳琳. 基于孪生网络的分类器输出重复性优化方法[J]. 计算机工程, 2025, 51(1): 118-127.
[4]	杨红菊, 吉昌. 学习驱动的图像压缩算法研究[J]. 计算机工程, 2025, 51(1): 190-197.
[5]	王晓路, 汶建荣. 基于运动-时间感知的人体动作识别方法[J]. 计算机工程, 2025, 51(1): 216-224.
[6]	火久元, 苏泓瑞, 武泽宇, 王婷娟. 基于改进YOLOv8的道路交通小目标车辆检测算法[J]. 计算机工程, 2025, 51(1): 246-257.
[7]	郑雅洲, 刘万平, 黄东. 一种基于注意力机制的BERT-CNN-GRU检测方法[J]. 计算机工程, 2025, 51(1): 258-268.
[8]	王骞, 张俊华, 王泽彤, 李博. X2S-Net:基于双平面X线片的脊柱三维重建[J]. 计算机工程, 2025, 51(1): 277-286.
[9]	易鹏, 杨晔, 严仕嘉. 基于MPCNN模型的sEMG快速迁移学习的手势识别应用研究[J]. 计算机工程, 2025, 51(1): 304-311.
[10]	刘兆伟, 方艳红, 郑明宇, 锁斌. 基于注意力机制与多任务的肺部疾病诊断方法[J]. 计算机工程, 2025, 51(1): 332-342.
[11]	魏嵬, 丁香香, 郭梦星, 杨钊, 刘辉. 文本相似度计算方法综述[J]. 计算机工程, 2024, 50(9): 18-32.
[12]	王志浩, 钱沄涛. 基于Swin Transformer的双流遥感图像时空融合超分辨率重建[J]. 计算机工程, 2024, 50(9): 33-45.
[13]	李俊俊, 董建刚, 李坤. 基于Kubernetes的集群节能策略研究[J]. 计算机工程, 2024, 50(9): 82-91.
[14]	张鲁, 田春伟, 宋焕生, 刘侍刚. 用于低剂量CT图像去噪的多级双树复小波网络[J]. 计算机工程, 2024, 50(9): 266-275.
[15]	朱凯, 李理, 张彤, 江晟, 别一鸣. 基于Transformer的多阶段运动模糊图像修复网络[J]. 计算机工程, 2024, 50(9): 276-285.

选择文件类型/文献管理软件名称

选择包含的内容