FACE LIVENESS DETECTION FOR AUTHENTICATION SYSTEMS USING CONVOLUTIONAL NEURAL NETWORKS
Face Anti-Spoofing is profoundly fundamental in the two scholastics and modern fields. Un-approved individuals are attempting to get confirmed through face introduction assaults (PAs, for example, a printed face photo, showing recordings on computerized gadgets, or a 3D veil assault). Along these lines, face introduction assault recognition (facial anti- spoofing) is required, which is the errand of forestalling bogus facial check by utilizing a photograph, video, veil, or an alternate substitute for an approved individual's face. The multimodal (RGB, profundity, and IR) technique dependent on CNN is proposed in this work for anti-spoofing of face for validation. The proposed technique demonstrated preferable presentation over the single model classifiers. Even though the multi-model demonstrated improved Performance, Feather-Net will present A/B network to decrease the unpredictability. This design utilized the combination technique in the Face Anti-mocking Attack identification and accomplished preferred outcomes over multi-model ways.