Siamese Neural Networks for One-Shot Face Recognition

A face recognition system is a piece of technology that can assess one’s face in an input photo or frames of a video against a collection of faces. By identifying and quantifying face characteristics from a picture, such a system may be used to identify individuals. Due to its numerous uses in areas including healthcare, security, advertising, identity authentication, monitoring, etc., it has caught the interest of many researchers. “One shot” refers to learning from an individual training item. A one-shot face recognition system tries to mimic the ability to draw lessons from a single experience. This approach is highly useful when the dataset is limited. We developed a face recognition system that is built with a deep neural network called Siamese Neural Network which adapts one-shot learning algorithms. The system was tested with 1 shot, 2 shot and 5 shot having 90 classes and 150 classes.