<![CDATA[This study systematically reviews the application of Convolutional Neural Networks (CNN) for gender classification using fingerprint images as a biometric identifier. CNN demonstrate a strong ability to automatically extract ridge and texture features that differentiate male and female fingerprints. Using the PRISMA 2020 framework, a Systematic Literature Review (SLR) was conducted on 15 Scopus-indexed studies published between 2020 with 2025, focusing on CNN-based fingerprint gender recognition. The review reveals that CNN models achieved accuracy ranging from 85% to 99.97%, depending on the network architecture, dataset size, and training strategy. Architectures such as VGG16, GoogleLeNet, AlexNet, EfficientNetB0, and Hybrid Model CNN–LSTM or CNN–SVM performed best, especially when enhanced with data augmentation and transfer learning. Interpretability methods such as Grad-CAM improve model transparency by visualizing fingerprint regions influencing gender prediction. Although the trend in accuracy slightly declined after 2023 due to dataset diversity and overfitting in deeper models, CNN remains the dominant and most reliable approach in biometric analysis.]]>
Copyrights © 2026
