Cartoon character identity recognition method based on generative adversarial network

Abstract

The invention discloses a cartoon character identity recognition method based on a generative adversarial network, and the method comprises the steps: obtaining a real pedestrian image and a cartoon pedestrian image, and building a retrieval library containing a real pedestrian and a query library containing a cartoon pedestrian; constructing a face detection alignment model, and performing face detection alignment on the pictures in the retrieval library and the query library; constructing a cartoon generation model, and converting real pedestrians and faces into corresponding cartoon pictures; constructing a cartoon character identity recognition model, extracting fusion features of pictures in the retrieval library and the query library, and calculating similarity scores between fusionfeatures of cartoon pedestrians and real pedestrians; calculating similarity scores between the cartoon pedestrian fusion features, reordering the similarity scores between the cartoon pedestrians andthe real pedestrians by utilizing the similarity scores between the cartoon pedestrians, and setting a threshold value to obtain real pedestrians corresponding to the cartoon pedestrians in the retrieval library in the query library. The method has the advantages of high precision and high speed for cartoon character identity recognition.

Description

technical field [0001] The present invention relates to the field of computer vision, and more specifically, relates to a method for recognizing comic characters based on generative adversarial networks. Background technique [0002] Face recognition has been a key research problem in the field of computer vision for the past few decades. In recent years, with the rapid development of technology, especially with the rapid development of deep learning, the deep face recognition model has reached or even surpassed the recognition level of humans on some data sets, for example, the most commonly used LFW in face recognition On the data set, the existing face recognition algorithm can achieve an accuracy rate of more than 99%. However, most of the existing face recognition algorithms are researched around real face images, and there are few special algorithms to study the subject of comic face recognition, so as to identify the identity of comic characters. [0003] Face recog...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the deficiencies of existing cartoon character identification methods and the shortcomings based on the single feature of traditional face recognition algorithms, the present invention proposes a cartoon character identification method based on generative confrontation network, which uses cartoon generation, cartoon face and Pedestrian feature fusion, caricature picture style classification and reordering strategies can improve the accuracy of real pedestrians in the search library for caricature pedestrians in the query database

Images