A Face Recognition Method for Real-time Updating Cosine Angle Loss Function Parameters

Abstract

The invention proposes a face recognition method for updating cosine angle loss function parameters in real time, which belongs to the field of face recognition in computer vision. The method includes: (1) collecting face images, classifying the face images as individuals, and performing data labeling on each face image; (2) performing image preprocessing on the face images to obtain a face image dataset; (3) Initialize the cosine value amplification scale and cosine angle interval of the cosine angle loss function; (4) Send the image data set into the convolutional neural network, and calculate and update the cosine value amplification scale and cosine angle interval in real time until the completion For the training of the convolutional neural network, (5) input the face image that needs to be compared and judged into the trained convolutional neural network, and output the face feature vector for face recognition and matching. The face recognition method of the present invention has the characteristics of fast training convergence speed and high recognition accuracy.

Description

technical field [0001] The invention belongs to the field of face recognition in computer vision, and in particular relates to a face recognition method for updating cosine angle loss function parameters in real time based on an adaptive learning mechanism. Background technique [0002] With the development of computer vision technology and the increasing demand for intelligent security and e-commerce, face recognition technology has become the most widely used field of artificial intelligence application products. The core of face recognition technology using deep learning method is to abstract the features of face images through convolutional neural network (CNN), which is used to calculate the similarity between face images, and then realize the function of face recognition. [0003] In the practical application of face recognition, it is often affected by factors such as camera imaging, lighting, and facial occlusion, resulting in a decline in recognition ability, thereb...

AI Technical Summary

Problems solved by technology

However, this method is highly random, and with the iterative transformation of the training data, the best value obtained before may not be applicable to the new data and network

Images