Skip to content

YuLiu-LY/Face-Recognition

Repository files navigation

Face recoginition project

This repository is a project for face recognition.

Environment Setup

We provide all environment configurations in requirements.txt. To install all packages, you can create a conda environment and install the packages as follows:

conda create -n face python=3.8
conda activate face
conda install pytorch==1.10.1 torchvision==0.11.2 torchaudio==0.10.1 cudatoolkit=11.3 -c pytorch -c conda-forge
pip install -r requirements.txt
pip install dlib
pip install face_recognition

In our experiments, we used NVIDIA CUDA 11.3 on Ubuntu 20.04. Similar CUDA version should also be acceptable with corresponding version control for torch and torchvision.

Data processing

Before you start training, you need to run data_process.py to process the raw data. Remember to change the DATA_ROOT to your own path.

Training

To train the model from scratch, we provide the following script:

$ chmod +x train.sh
$ ./train.sh

Remember to change the data_root and log_path to your own path.

Reload ckpts & test_only

To reload checkpoints and only run inference, we provide the following script:

$ chmod +x test.sh
$ ./test.sh

We provide two checkpoints in ckpt folder: train_val.ckpt is trained on the whole dataset, and train.ckpt is trained on the training dataset. You can change the test_ckpt_path in test.sh to reload the checkpoints. We provide all checkpoints at this link. We also provide two types of evaluation: test and val. You can change the action in test.sh to choose the evaluation type. Remember to change the data_root and log_path to your own path.

Test face recognition

To test face recognition, you can run fr_acc.py to get the accuracy of face recognition. Remember to change the DATA_ROOT to your own path.

Acknowledgement

This code used resources from BO-QSA. We thank the authors for open-sourcing their awesome projects.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published