Face Clustering on Image Repository Using Convolutional Neural Network

Face clustering is highly related to face recognition. For performing face recognition we are applying unsupervised learning where we have images of faces we want to cluster, to achieve Face detection and clustering application in real-time on an image repository. The entire face clustering is divided into two modules. The first module is face detection and the second module is clustering. For clustering we are using DBSCAN (Density-Based Spatial Clustering), since the DBSCAN algorithm naturally handles outliers, marking them as such if they fall in low-density regions where their nearest neighbors are far away. To detect the face in the image and to retrieve the corresponding embedding (128-dimensional features), we use face recognition API that internally uses CNN (Convolutional Neural Networks) for recognizing faces and extracting embedding’s from those recognized faces.