Method for improving performance of face recognition model under glasses wearing condition

Abstract

The invention discloses a method for improving performance of a face recognition model under a glasses wearing condition. The method comprises the following steps: extending existing glasses non-wearing face training data as glasses wearing face training data by virtue of an automatic face image glasses wearing algorithm; training by utilizing the extended glasses wearing face training data, thereby obtaining the face recognition model. By increasing glasses for a face image without wearing glasses, the existing glasses non-wearing training data can be rapidly and conveniently extended into the glasses wearing training data, and the scale and diversity of the training data can be improved. Compared with a method for newly building glasses wearing face training data, the method disclosed bythe invention is low in cost, convenient, rapid and obvious in effect, and lots of manual labor and financial cost expense can be saved. Meanwhile, the face recognition model is trained by virtue ofextended training samples, so that the face recognition model has excellent interference resistance and recognition effect on the glasses wearing face, and the overall recognition accuracy is greatlyimproved.

Description

technical field [0001] The invention belongs to the field of computer vision, relates to a face detection and recognition method, in particular to a method for improving the performance of a face recognition model under the condition of wearing glasses. Background technique [0002] In order to improve the recognition effect of the face recognition algorithm, it is usually necessary to use a large amount of training data for training in order to obtain a face recognition model with excellent performance. Under the premise of the same model structure, the scale and diversity of training data will have a decisive impact on the final performance of the model. In practical applications, face recognition technology usually faces the following problems: First, most of the existing training data are training data without glasses, and the face recognition model obtained through the training data is less effective for face recognition with glasses. At the same time, if a new large-s...

AI Technical Summary

Problems solved by technology

In practical applications, face recognition technology usually faces the following problems: First, most of the existing training data is training data without glasses, and the face recognition model obtained through the training data is less effective for face recognition with glasses. At the same time, if a new large-scale face training data is created, it will not only consume a lot of manpower and fina

Images