Skip to content

Latest commit

 

History

History
41 lines (29 loc) · 1.68 KB

README.md

File metadata and controls

41 lines (29 loc) · 1.68 KB
Raw

RECBOT: Experiments on Face detection and recognition

Uses HAR cascades for fast good-enough face detection. Uses SFace for face recognition via cosine similarity against model computed features of samples.

Remember:

  • Face Detection: is there a face in the image?
  • Face Recognition: whose face is in the image?

Dependencies

  • Install/Compile libopencv-dev and opencv-data
  • Install cmake, make, and g++

Will use -march=native in order to enable SSE, AVX, et al. where possible.

Using

  • make prepare will prepare cmake directories and download model and it's LICENSE.
  • make build will build everything
  • make sample will run the sampler to capture 10 samples of a face using the computer camera and crop it to the bounding box found by the face detector.
  • make recon will run the recognizer and output "PASS" or "FAIL", with exit codes 0 or -1 respectively.

Binaries can be found on _build/.

Notes:

Choosing the camera

You can change the camera by changing the index (0) on: #define CAMERA_INDEX 0 in both recbot_recon.cpp and recbot_samples.cpp. Use ls /dev/video* to list, and note that not all devices are actual cameras.

Simple Detection Example

Under ./examples/ you can check a simple face detection example using HAR Cascades.

Troubleshooting:

If vscode intellisense is not working add "/usr/include/**" or equivalent to includePath in the C/C++ properties (c_cpp_properties.json).

TODO:

  • Command line argument parsing
  • Make this work with PAM for authentication
  • Ensure memory is not leaking, because right now this runs as a one shot POC and I didn't care.