Setup instructions macOS-only

Assumptions

• you have brew set up
• you have zsh set up
• you have xcode command line tools; if not run xcode-select --install

Installation

• Run ./install.sh
• Run source ~/.zshrc
• Includes setting up miniconda and torch binaries
• In some cases, you may need to perform these two additional steps:
  • brew install libpng
  • xcode-select --install

Verify and run

• Enter the virtual env: conda activate xconf
• VIDEO CAPTURE: python 1_capture_video.py
• DETECT FACES: python 2_detect_faces.py
• ALIGN FACES: python 3_align_faces.py - opens in two windows
• FIND EMBEDDINGS: python 4_find_embeddings.py
• TRAIN
  • Create a folder in TrainingData with your name
  • Place your selfies in the folder
  • Run python 5_train_classifier.py
• TEST
  • python 6_test_classifier.py [--mode test|train]

Uninstall

• Run ./clean.sh
• Remove conda and torch references from ~/.zshrc manually

References

• https://medium.com/@ageitgey/machine-learning-is-fun-part-4-modern-face-recognition-with-deep-learning-c3cffc121d78

Related Work

• https://github.com/twcbe/known_face_detector