Face detection & Emotion recognition

---

Table of Contents

• Pre-requisites
• Quick Start
• Usage
  • Dataset
  • Face Detection
  • Pre-trained Model
• Sample Outputs
• References

Pre-requisites

• argparse

  pip install argparse

• Keras

  pip install keras

• opencv-python

  pip install opencv-python

• opencv-contrib-python

  pip install opencv-contrib-python

• Numpy

  pip install numpy

• facenet-pytorch

  pip install facenet-pytorch

Quick Start

• Clone this repository

$ git clone https://github.com/aiis-research/Face-detection

• Run the demo:

image input

$ python detect_demo.py --image data/image/image.jpg

video input

$ python detect_demo.py --video data/video/video.mp4

webcam

$ python detect_demo.py --src 0

Usage

Dataset

We use pre-trained model trained by FER2013 dataset(See here). The dataset consists of visual data of facial expression labeled with 7 emotion categories.

Face Detection

We used MTCNN(See here) as a facial recognition model to detect emotions.

Pre-trained Model

In order to inference the model, we used pre-trained model XCEPTION(See here) developed by Google(2017).

Sample Outputs

sample image output:

sample video output:

References

1. Tim Esler's facenet-pytorch repo: https://github.com/timesler/facenet-pytorch

2. Octavio Arriaga's pre-trained model repo: https://github.com/oarriaga/face_classification

3. K. Zhang, Z. Zhang, Z. Li and Y. Qiao. Joint Face Detection and Alignment Using Multitask Cascaded Convolutional Networks, IEEE Signal Processing Letters, 2016. PDF

4. F. Chollet. Xception: Deep Learning with Depthwise Separable Convolutions, 2017. PDF