V.A. Ivanov, A.A. Smirnov, K.D. Galev
Statement of the problem. Identification of age-related changes in persons requires the creation of many illegal acts against minors. To solve the problems of detecting age-related changes, the article presents a basic method based on deep neural networks.
Target. Consider the process of neural network identification of age-related changes based on features, extractable modules of «attention» and the module of «conditional determination of identification», as well as optimization of neural network weights.
Results. A model of the process of forming a feature space for the tasks of detecting age-related changes in human faces based on the accumulated database of color images of people at different ages is presented.
Practical significance. The approaches used in the study to the formation of a feature space for the detection of age-related changes in faces, the creation of a detection model based on informative features, and the optimization of the parameters of the resulting model can be applied to create biometric detection subsystems in order to solve the problem of preventing the forgery of identity documents.
Ivanov V.A., Smirnov A.A., Galev K.D. An age-invariant face recognition model. Radiotekhnika. 2023. V. 87. № 2. P. 47−52.
DOI: https://doi.org/10.18127/j00338486-202302-07 (In Russian)
- Hao Wang, Dihong Gong, Zhifeng Li, Wei Liu. Decorrelated adversarial learning for age-invariant face recognition. In IEEE Conf. Comput. Vis. Pattern Recog. 2019. P. 3527–3536.
- Jiankang Deng, Jia Guo, Niannan Xue, Stefanos Zafeiriou. Arcface: Additive angular margin loss for deep face recognition. In IEEE Conf. Comput. Vis. Pattern Recog. 2019. P. 4690–4699.