This is a face recognition framework based on PyTorch with convenient training, evaluation and feature extraction functions. It is originally a multi-task face recognition framework for our accpeted ECCV 2018 paper, "Consensus-Driven Propagation in Massive Unlabeled Data for Face Recognition". However, it is also a common framework for face recognition. You can freely customize your experiments with your data and configurations with it.

Xiaohang Zhan, Ziwei Liu, Junjie Yan, Dahua Lin, Chen Change Loy, "Consensus-Driven Propagation in Massive Unlabeled Data for Face Recognition", ECCV 2018

Project Page: link

Different datasets have different identity (category) sets. We do not know the intersection between them. Hence instead of merging identity sets of different datasets, regarding them as different tasks is an effective alternative way.

Framework: Multi-task, Single Task

Loss: Softmax Loss, ArcFace

Backbone CNN: ResNet, DenseNet, Inception, InceptionResNet, NASNet, VGG

Benchmarks: Megaface (FaceScrub), IJB-A, LFW, CFP-FF, CFP-FP, AgeDB-30, calfw, cplfw

Data aug: flip, scale, translation

Online testing and visualization with Tensorboard.

1. Clone the project.

   git clone [email protected]:XiaohangZhan/face_recognition_framework.git
   cd face_recognition_framework
   

2. Dependency.

   python>=3.6, tensorboardX, pytorch>=0.3.1, mxnet, sklearn

3. Data Preparation.

   Download datasets from insightface into your data storage folder, e.g., ~/data/face_recognition/. Taking CASIA-Webface for example:

   cd ~/data/face_recognition/
   unzip faces_CASIA_112x112.zip
   cd - # back to the repo root
   python tools/convert_data.py -r ~/data/face_recognition/faces_webface_112x112 -o ~/data/face_recognition/faces_webface_112x112 # convert mxnet records into images
   ln -s ~/data/face_recognition/faces_webface_112x112 data/webface

   Optionally, if you want to test on MegaFace. Download testing set from here into your data storage folder, e.g., ~/data/face_recognition/. Then:

   cd ~/data/face_recognition/
   mkdir -p megaface_test/raw
   cd megaface_test/raw
   mv ../../megaface_testpack_v1.0.zip .
   unzip -q megaface_testpack_v1.0.zip
   cd $THIS_REPO # back to the repo root
   ln -s  ~/data/face_recognition/megaface_test data/megaface_test

   Next, download MegaFace lists from here into ~/data/face_recognition/megaface_test/. Finally, the folder data/megaface_test/ looks like:

   data
     ├── megaface_test
       ├── concat_list.txt
       ├── facescrub3530
       ├── megaface_distractor
       ├── raw
   

4. Training.

   sh experiments/webface/res50-bs64-sz224-ep35/train.sh
   

5. Monitoring.

   tensorboard --logdir experiments
   

6. Resume training.

   sh experiments/webface/res50-bs64-sz224-ep35/resume.sh 10 # e.g., resume from epoch 10
   

7. Evalution.

   sh experiments/webface/res50-bs64-sz224-ep35/evaluation.sh 35 # e.g., evaluate epoch 35
   

8. Feature extraction.

   Firstly, specify the data_name, data_root and data_list under extract_info in the config file. The data_list is a txt file containing an image relative filename in each line. Then execute:

   # e.g., extract features with epoch 35 model.
   # The feature file is stored in checkpoints/ckpt_epoch_35_[data_name].bin
   sh experiments/webface/res50-bs64-sz224-ep35/extract.sh 35 
   

• Trained using Webface

• Trained using MS1M

Note that the hyper-parameters are not adjusted to optimal. Hence, they are not the state-of-the-art face recognition models. You may download those pre-trained models here.

If you find this code useful in your research, please cite:

@inproceedings{zhan2018consensus,
  title={Consensus-Driven Propagation in Massive Unlabeled Data for Face Recognition},
  author={Zhan, Xiaohang and Liu, Ziwei and Yan, Junjie and Lin, Dahua and Change Loy, Chen},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  pages={568--583},
  year={2018}
}

1. Implement distributed training.
2. Adjust hyper-parameters.
3. Multi-task experiments.