MTCNN: Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks

---

This repository is the implementation of MTCNN in TF2. It is written from scratch, using as a reference the implementation of MTCNN from David Sandberg Project: facenet.

Inference

Just download the repository and then do this

$ python main.py

The detection results contains three part:

• Face classification: the probability produced by the network that indi- cates a sample being a face
• Bounding box regression: the bounding boxes’ left top, height, and width.
• Facial landmark localization: There are five facial landmarks, including left eye, right eye, nose, left mouth corner, and right mouth corner

Citation

@Github_Project{TensorFlow2.0-Examples,
  author       = YunYang1994,
  email        = www.dreameryangyun@sjtu.edu.cn,
  title        = "MTCNN: Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks",
  url          = https://github.com/YunYang1994/TensorFlow2.0-Examples,
  year         = 2019,
}