Face recognition with Deepstream

This is a face detection and recognition demo pipeline build on Deepstream.

Requirements

• Deepstream 6.0+
• GStreamer 1.14.5+
• Cuda 11.4+
• NVIDIA driver 470.63.01+
• TensorRT 8+
• Python 3.6+
• Opencv

Follow deepstream official doc to install dependencies.

Deepstream docker is more recommended.

Pretrained

Please refer to links below for pretrained models and serialized TensorRT engine. Or download from Google driver.

• yolov5
• retinaface
• retinaface
• arcface

Alignment

there should be a face alignment before arcface. Use a gst-nvinfer-custom to preprocess the tensor-meta of retinaface.

Demo custom-gst-nvinfer

Usage

1 - put the feature file to data/known_faces

2 - compile gst-nvinfer-custom if do alignment, care that the post-process should also be modified if use gst-nvinfer-custom

3 - modify config file under config folder

4 - run python3 main.py

References

• wang-xinyu/tensorrtx
• NVIDIA-AI-IOT/deepstream_python_apps
• biubug6/Pytorch_Retinaface