RetinaFace (RetinaFace: Single-stage Dense Face Localisation in the Wild, published in 2019) reimplemented in Tensorflow 2.0, with pretrained weights available. Resnet50 backbone.

Original paper -> arXiv
Original Mxnet implementation -> Insightface

1. Installation
2. Usage
3. Benchmark
4. Evaluation
5. Acknowledgements

example output :

To install dependencies, run :

pip install -r requirements.txt

Then build the rcnn module by running :

make

Run :

python detect.py --sample_img="./sample-images/WC_FR.jpeg"

Python usage :

from src.retinafacetf2.retinaface import RetinaFace
import cv2

detector = RetinaFace(False, 0.4)
img = cv2.imread("./sample-images/WC_FR.jpeg")
faces, landmarks = detector.detect(img, 0.9)

mAP result values on the WIDERFACE validation dataset:

In order to verify the models accuracy on the WiderFace dataset:

• Run the model on the dataset and generate text files as results

python eval_widerface --widerface_data_dir = "/data/WIDER_test/images" --save_folder="./WiderFace-Evaluation/results/"

• Evaluate the results

cd ./WiderFace-Evaluation
python setup.py build_ext --inplace
python evaluation.py -p ./results_val/ -g ./ground_truth/

This work is largely based on the original implementation by the amazing insightface team
Evaluation on widerface done with the Widerface-Evaluation repo
If you use this repo, please reference the original work :

@inproceedings{Deng2020CVPR,
title = {RetinaFace: Single-Shot Multi-Level Face Localisation in the Wild},
author = {Deng, Jiankang and Guo, Jia and Ververas, Evangelos and Kotsia, Irene and Zafeiriou, Stefanos},
booktitle = {CVPR},
year = {2020}
}