Multi Stream CNN for biometric recognition- Deep Learning projects-Matlab
In this project,biometric recognition is done thru Convolutional neural network.Nowadays biometric recognition is most important task to identify a person.
We propose a deep multi stream CNN network to fuse multiple modalities (face, iris, and fingerprint) for person identification. The proposed deep multi stream algorithm consists of multiple streams of modality-specific Convolutional Neural Networks (CNNs), which are jointly optimized at multiple feature abstraction levels. Multiple features are extracted at several different convolutional layers from each modality-specific CNN for joint feature fusion, optimization, and classification. Features extracted at different convolutional layers of a modality-specific CNN represent the input at several different levels of abstract representations.
Experimental results,show that our deep multi-stream CNNs with biometric recognition at several different feature level abstraction can significantly outperform the unimodal representation accuracy. Experimental results on the datasets has shown significant capability for identification biometric system. The proposed work can offer an accurate and efficient matching compared with other system based on unimodal, bimodal, multimodal characteristics.