Approximate Rank-Order Clustering (AROC)

Implementation of Approximate Rank-Order Clustering (AROC) algorithm in Clustering Millions of Faces by Identity. Features used in the implementation can be found at face_verification_experiment by AlfredXiangWu. Features extracted by other models can be used too.

How to run

mkdir data

# Download the LFW features from [face_verification_experiment by AlfredXiangWu]
wget https://github.com/AlfredXiangWu/face_verification_experiment/raw/master/results/LightenedCNN_C_lfw.mat -P data

# Perform the clustering
python main.py