Face Clustering Method Based on GCN
Ran Ma · 2024
Face clustering is an important research module in the field of image clustering. Face clustering generally refers to recognition and classification of facial image, is widely used in fields such as Pattern recognition, Human-Computer interaction, and Information security. In recent years, face clustering methods based on Graph convolutional network (GCN) have made significant progress, but there are still problems of generating too many subgraphs and high subgraph redundancy. In response to these existing problems, proposing a clustering method based on graph convolutional network with improved subgraph generation algorithm and innovative subgraph pruning algorithm. Experimental results show that this method not only effectively solves the problems of generating too many subgraphs and high redundancy of subgraphs in face clustering process, but also reduces random error and accelerates the convergence of graph convolutional network, thereby improving model performance. While proposing this method, this article deeply considers the scalability and compatibility of the module, making further research based on this method helpful in promoting the development of adaptive large-scale face recognition system.