Research-internship: face clustering

In this repository multiple face clustering methods are implemented. These methods work on top of the facenet implementation provided by https://github.com/arunmandal53/facematch.

storeFaces.py

This file extracts faces from images stored in the folder /images. The faces containing the file name, id and embedding are stored in a Json file called data.json.

distanceTable.py

Has a function getDistanceTable that reads the faces in the Json file and returns a distance matrix containing all distances between faces in the file.

thresholdClustering.py

Clusters the faces from the image folder to a cluster folder. If a difference between faces is under a certain threshold the faces are stored in the same cluster.

kmeans.py

Simple implementation of Kmeans algorithm on the distance matrix.

meanshift.py

Simple implementation of the mean shift clustering algorithm on the distance matrix.

hierarchicalClustering.py

Simple implementation of a dendrogram calculation on the distance matrix.