Ensemble Learning Framework for Blood Group Identification Using Fingerprint Patterns

This study explores fingerprint-based blood group prediction using deep learning models and ensemble techniques. We compared ResNet, DenseNet, EfficientNet, and a Stacked XG-Boost classifier. The Stacked XGBoost model achieved the highest accuracy of 0.8959 on loaded test data. Feature importance analysis identified key fingerprint features, demonstrating the potential of ensemble methods for accurate biometric classification.