Academic Innovations

The research team explored digital biometric parameters such as skin types and moisture content, crucial fornoptimizing fingerprint sensor functionality. Public biometric data, particularly from LivDet 2015, was extensively analyzed to train and validate AI models, ensuring robust performance under diverse conditions. AI models, including Convolutional Neural Networks (CNNs) and Transformer architectures, were evaluated for their efficacy in data pre-processing and feature extraction, enhancing the accuracy of subsequent algorithms. Efforts also started for optimizing AI/ML models for reduced latency and memory usage, implementing efficient architectures like MobileNet and EfficientNet to enable real-time processing capabilities.