Barefoot footprint image personal recognition method based on multi-scale depth features

The invention discloses a barefoot footprint image personal recognition method based on multi-scale depth features, which relates to the technical field of barefoot image recognition and deep learning. The method comprises the steps of obtaining a barefoot footprint image, and constructing a large-scale barefoot footprint image data set, dividing the barefoot footprint image data set into a training set, a verification set and a test set, preprocessing the barefoot footprint image, performing data enhancement on the pre-processed barefoot footprint image, constructing a deep neural network for barefoot feature extraction, constructing a loss function, and training the deep neural network by using a metric learning method to obtain a barefoot image personal recognition model, and using the barefoot image personal recognition model to extract barefoot features for personal recognition. The method can cover barefoot features from the global to the local at the same time, so that when the method is applied to a p