Face Detection @ 500-1000 FPS

Reimplementation of RetinaFace, faster and stronger.

Getting Start

  • Install gstreamer for reading videos

    sudo apt-get install libgstreamer1.0-0 gstreamer1.0-plugins-base gstreamer1.0-plugins-good gstreamer1.0-plugins-bad gstreamer1.0-plugins-ugly gstreamer1.0-libav gstreamer1.0-doc gstreamer1.0-tools gstreamer1.0-x gstreamer1.0-alsa gstreamer1.0-gl gstreamer1.0-gtk3 gstreamer1.0-qt5 gstreamer1.0-pulseaudio

  • Testing on usb camera 0   (Be Careful About ! and |)

    gst-launch-1.0 -q v4l2src device=/dev/video0 ! video/x-raw, width=640, height=480 ! videoconvert ! video/x-raw, format=BGR ! fdsink | python3 face_detector.py

PR on wilder face test dataset

faster-mobile-retinaface

Speed on GTX1660ti and Jetson-nano

  • Why use Faster-RetinaFace ?

    Plan Inference Postprocess Throughput Capacity (FPS)
    9750HQ+1660TI 0.9ms 1.5ms 500~1000
    Jetson-Nano 4.6ms 11.4ms 80~200

    If the queue is bigger enough, the throughput capacity can reach the highest.

faster-mobile-retinafacev

Citation

@inproceedings{deng2019retinaface,
title={RetinaFace: Single-stage Dense Face Localisation in the Wild},
author={Deng, Jiankang and Guo, Jia and Yuxiang, Zhou and Jinke Yu and Irene Kotsia and Zafeiriou, Stefanos},
booktitle={arxiv},
year={2019}
}

GitHub