Skip to content

Latest commit

 

History

History
39 lines (33 loc) · 1.02 KB

README.md

File metadata and controls

39 lines (33 loc) · 1.02 KB
Raw

Face ID with Jetson Nano

Using MTCNN and MobileFacenet architecture with ArcFace with C++

Required:

  • Libtorch
  • OpenCV 3.x.x
  • OpenBlas
  • Cuda/Cudnn
  • TensorRT

Installation:

For Pytorch. Installation follow link:

https://forums.developer.nvidia.com/t/pytorch-for-jetson-version-1-7-0-now-available/72048

Next change the path {user} with your user name in CMakeList.txt. In this repository, I change the name of CMakeList for Jetson Nano to CMakeListJetson.txt

For MTCNN, follow link:

https://github.com/AlphaQi/MTCNN-light

Face Classifier:

Using LibSVM for classification. LibSVM can train on multi-class and produce probability for each class - which can be used for detecting Unknown Identification. For more detail, see:

https://github.com/cjlin1/libsvm

I have edited code in LibSVM so when processing, our program doesn't need to write data to txt file. Make it easier and faster to process.

Compile and run:

cmake .
make
./main

Hope this can help you guys