Age estimation using biometric traits has gained significant attention in recent years due to its applications in security, surveillance, access control, and digital identity verification systems. However, existing approaches that rely on a single facial or ocular modality often suffer from reduced accuracy under variations such as illumination changes, occlusion, and cosmetic alterations. To overcome these limitations, this paper proposes a hybrid deep learning framework that integrates iris and periocular features for robust human age estimation. The proposed architecture adopts a dual-branch convolutional neural network structure, where the iris region is processed using a deep residual network to capture fine-grained micro-textural patterns, while the periocular region is analyzed using a convolutional network to extract wrinkle-based and skin-texture aging cues. Feature-level fusion is performed by concatenating the extracted representations from both branches, forming a comprehensive hybrid feature vector. This fused representation is further passed through fully connected layers to predict the chronological age. The integration of internal iris characteristics and external periocular features enhances discriminative capability and improves resistance to environmental variations. Experimental results demonstrate that the proposed multimodal fusion framework significantly outperforms single-modality approaches, achieving an overall accuracy of 97% in age classification. The findings confirm that combining complementary ocular modalities leads to improved robustness, reliability, and generalization performance, making the proposed system suitable for real-world biometric age estimation applications.
R et al. (Fri,) studied this question.