A CNN Based Embedded System for Improved Face Recognition

Face recognition with the help of deep learning and machine learning has become a key research objective of many researchers and academicians. The techniques and algorithms used earlier were restricted to preserve the information as much as possible and reduce information redundancy. This face recognition model involves gathering the substitute features and inserts the feature vectors as classifiers; these classifiers are widely used in Neural Networks. After the advancements in deep learning technology convolution, neural networks were used to imbibe a large number of face data. In practice, this neural network model was usually large and consists of more parameters and yields better performance results, also while considering different constraints like recognizance accuracy, speed of processing in data, then the deep learning method, using the neural network based on the triplet loss, and using the compression quantization method to optimize the face recognition on the embedded device is used to be designed. This paper aims at the design and construction of such an embedded device. This work involves CNN - based deep learning, Raspberry - pi, IoT, and Python Language.