hosituan / clockon-clockoff-face-recognition

Face Recognize application. Using FaceNet and CoreML

clockon-clockoff-face-recognition

This is a Face Recognition Application - HR Check-in on iOS Device.

References

• Pre-train Model: Facenet https://github.com/davidsandberg/facenet
• Model for TensorFlow Lite: https://drive.google.com/file/d/1WjZ-rX-Jq0U08U5VefQ7XSC5Dc6RPitI/view
• Dataset: VGGFace2 https://drive.google.com/file/d/1EXPBSXwTaqrSC0OhUdXNmKSh9qJUQ55-/view?usp=drive_open
• Architecture: InceptionResnetV1 https://github.com/davidsandberg/facenet/blob/master/src/models/inception_resnet_v1.py
• Algorithms and Libraries:
  • Swift for TensorFlow: https://www.tensorflow.org/swift
  • Swift K-means Clustering https://github.com/raywenderlich/swift-algorithm-club/tree/master/K-Means
  • Swift k-NN https://github.com/mmahler2/Swift-DTW-KNN
  • Apple Vision: https://developer.apple.com/documentation/vision
  • Apple CoreML: https://developer.apple.com/documentation/coreml
  • Apple TuriCreate: https://github.com/apple/turicreate

Supported Platforms

• iOS 12.0 or later.
• Xcode 11 or later, Swift 5.

Demo

Performance

• Test device: iPhone X, iOS 14.2

• Number of people: 50 persons.

• Time taken: ~0.12 seconds.

• Accuracy: >= 90%

Screen

• Recognize Screen:

  

• Unknown Person:

  

• Predict Image Screen (two people):

  

• Predict Image Screen, Time Taken (1 person):

  

• Time Logs Screen:

  

Usage

pod install

pod update

Author

Hồ Sĩ Tuấn - iOS Developer

Contact:

• Email: [e-mail]
• Phone: [phone]
• Country: VietNam
• Facebook: https://www.facebook.com/sytuann/
• Github: hosituanit

Contributor:

• Lâm Gia Khánh

Contributing

Issues and pull requests are welcome! Feel free to folk and fix.

Copyright

Please give me a star and reference link to my respository.