Deep Learning Approaches for Classification and Gender Prediction from Fingerprint Imagery: A Review

This review of literature in Fingerprint classification and Gender prediction not only catalogues the methodologies but also highlights the recurring trends across studies. As an example, though datasets such as SOCOFing, NIST and custom collected fingerprints are frequently used, it suffers from limited demographic diversity and imbalanced classes. Classical approaches rely on minutiae distribution and ridge density while Deep Learning approaches leverage Convolutional Neural Networks (CNN) to learn representations from fingerprint images for extracting features. The Machine Learning methodologies used in the studies range from Support Vector Machines (SVMs) to ensemble models. Metrics for Performance used are Accuracy, ROC-AUC, Recall, Precision and F1 Score with Deep Learning approaches outperforming at the cost of Computational complexity. The comparison of the literature underscores the key findings as the CNNs achieves superior accuracy for both classification and Gender prediction using fingerprints. However, the limitations include the Dataset imbalance, Less sample size and restricted demographic diversity. Finally, the consolidated recommendations of the review include Dataset generalizability to include diverse population groups and balanced age, gender distributions, Data augmentation leveraging Generative AI to mitigate imbalances, Multimodal biometric exploration for improved performance, Model generalization by benchmarking across multiple datasets.

[1]         A. Quinteros and D. Zabala-Blanco, “Fingerprint Classification Based on Multilayer Extreme Learning Machines,” Appl. Sci., vol. 15, no. 5, 2025, doi: 10.3390/app15052793. [2]         D. K. K, D. S. G. L, D. G. S. A, D. D. R, and D. K. R, “Optimized fingerprint crime detection using robust deformed convolutional neural network for 5G network secure smart cities,” Knowledge-Based Syst., vol. 317, pp. 113342–113342, 2025, doi: 10.1016/j.knosys.2025.113342. [3]         A. K. F. Shahad Hazim Mahmood, “An evaluation of convolutional neural network architectures: AlexNet, LetNet, VGG-16, and fingerprint-net for fingerprint classification,” 2025. [4]         A. Haleem, S. Ben Jabra, and W. Chainbi, “Biometric fingerprint identification in low-quality samples using a hybrid features extraction method with deep learning,” Evol. Syst., vol. 16, no. 3, p. 3, 2025, doi: 10.1007/s12530-025-09726-5. [5]         P. Choudhary, P. Pathak, and P. Gupta, “Assessing fingerprint image quality: an ensemble technique perspective,” Eng. Res. Express, vol. 7, no. 1, 2025, doi: 10.1088/2631-8695/adbb3c. [6]         A. Hussian, F. Murshed, M. N. Alandoli, and G. Aljafari, “A Hybrid Deep Learning Approach for Secure Biometric Authentication Using Fingerprint Data,” Computers, vol. 14, no. 5, pp. 1–19, 2025, doi: 10.3390/computers14050178. [7]         K. M. and J. P. M. S. A, J. PA, Sunitha Nandhini, Manimegalai R, “Automated Minutiae Extraction and Fingerprint Analysis Using Machine Learning Techniques,” in International Conference on Computational Innovations and Engineering Sustainability (ICCIES), 2025, pp. 1–9. [8]         D. Maiti, M. Basak, and D. Das, “Deep Learning Method for Multi-Attribute Analysis of Fingerprint Images,” Comput. Sci. J. Mold., vol. 32, no. 2, pp. 199–222, 2024, doi: 10.56415/csjm.v32.11. [9]         A. M. and N. E. B. A. Walhazi, Hajer, “A multi-classifier system for automatic fingerprint classification using transfer learning and majority voting,” Multimed. Tools Appl., pp. 6113–6136, 2024. [10]       M. B. Bhilavade, K. S. Shivaprakasha, M. R. Patil, and L. S. Admuthe, “Fingerprint Reconstruction: Approaches to Improve Fingerprint Images,” J. Wirel. Mob. Networks, Ubiquitous Comput. Dependable Appl., vol. 15, no. 1, pp. 75–87, 2024, doi: 10.58346/JOWUA.2024.I1.006.