This code is the implementation for the CVPR 2021 paper "Structure-Aware Face Clustering on a Large-Scale Graph with 10^7 Nodes", and the extended journal version "STAR-FC: Structure-Aware Face Clustering on Ultra-Large-Scale Graphs". 🌟🌟.

An extended version "STAR-FC: Structure-Aware Face Clustering on Ultra-Large-Scale Graphs" is proposed for face clustering on ultra-large-scale graphs with hierarchical GCN training, whcih can boost the face clustering performance from 91.97 to 93.21 in terms of pairwise F-score on standard partial MS1M within 312s!

The training and inference processes are as following:

For training, adjust the configuration in ./src/configs/cfg_gcn_ms1m_hierarchical.py, then run the algorithm as follows:

cd STAR-FC
sh scripts/train_gcn_ms1m_hierarchical.sh

For testing, adjust the configuration in ./src/configs/cfg_gcn_ms1m_hierarchical.py, then run the algorithm as follows:

cd STAR-FC
python test_dynamic.py

• Python = 3.6
• Pytorch = 1.2.0
• faiss

The hardware we used in this work is as follows:

• 24G TITAN RTX
• 48 core Intel Xeon CPU E5-2650-v4@2.20GHz processor

cd STAR-FC

Create a new folder for training data:

mkdir data

To run the code, please download the refined MS1M dataset and partition it into 10 splits, then construct the data directory as follows:

|——data
   |——features
      |——part0_train.bin
      |——part1_test.bin
      |——...
      |——part9_test.bin
   |——labels
      |——part0_train.meta
      |——part1_test.meta
      |——...
      |——part9_test.meta
   |——knns
      |——part0_train/faiss_k_80.npz
      |——part1_test/faiss_k_80.npz
      |——...
      |——part9_test/faiss_k_80.npz

We have used the data from: https://github.com/yl-1993/learn-to-cluster

Put the pretrained models Backbone.pth and Head.pth in the ./pretrained_model. Our trained models will come soon. Recently, some people ask for the pretrained model. I have't sorted out these models carefully (Maybe early June). However, to help research, I will release a model and you can found it in this link: https://cloud.tsinghua.edu.cn/d/cbd04a98b7d148dbae9e/, the password is: STAR-FC_CVPR.

Adjust the configuration in ./src/configs/cfg_gcn_ms1m.py, then run the algorithm as follows:

cd STAR-FC
sh scripts/train_gcn_ms1m.sh

Adjust the configuration in ./src/configs/cfg_gcn_ms1m.py, then run the algorithm as follows:

cd STAR-FC
python test_final.py

This code is based on the publicly available face clustering codebase https://github.com/yl-1993/learn-to-cluster.

Please cite the following paper if you use this repository in your reseach.

@inproceedings{shen2021starfc,
   author={Shen, Shuai and Li, Wanhua and Zhu, Zheng and Huang, Guan and Du, Dalong and Lu, Jiwen and Zhou, Jie},
   title={Structure-Aware Face Clustering on a Large-Scale Graph with 10^7 Nodes},
   booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
   year={2021}
}