现代信息科技2024，Vol.8Issue(11)：116-120,5.DOI:10.19850/j.cnki.2096-4706.2024.11.023
基于图卷积神经网络的人脸属性识别
Face Attribute Recognition Based on Graph Convolutional Neural Networks
摘要
Abstract
The research on multi-attribute recognition of facial images and the modeling of dependencies between multiple labels is a highly concerned research topic in the fields of Computer Vision and Machine Learning.A multi-label facial attribute recognition model based on Graph Convolutional Neural Networks is proposed to improve recognition efficiency by leveraging the dependency relationships between multiple labels.This model constructs a directed graph between facial attributes in a data-driven manner,and maps each attribute to the corresponding attribute classifier using a Graph Convolutional Neural Networks to model the dependency relationships between categories.The model has conducted in-depth analysis on key elements such as correlation matrix and feature matrix in Graph Convolutional Neural Networks,enabling it to handle multi-label facial attribute recognition problems.The experimental results show that the model performs well on the authoritative dataset CelebA for multi-label facial attribute recognition and can maintain a meaningful semantic structure.关键词
深度学习/人脸属性识别/图卷积神经网络/多标签分类Key words
Deep Learning/face attribute recognition/Graph Convolutional Neural Networks/multi-label classification分类
信息技术与安全科学引用本文复制引用
李名涵,刘科,昂寅..基于图卷积神经网络的人脸属性识别[J].现代信息科技,2024,8(11):116-120,5.基金项目
中南民族大学教研项目(JYX19062) （JYX19062）
