/emotionRecognition

Facial expression recognition system based on Improved Adaboost with Gabor Features

Primary LanguagePython

Emotion-Recognition

Open In Colab

Facial expression recognition system based on Improved Adaboost with Gabor Features.

Summary

  • A Simple 2-layer neural network Improved AdaBoost based facial expression recognition system with Gabor Features.
  • Average recognition rates in JAFFE, and CK+ is 98% and 97% respectively.
  • Supported Classes are Happy (0), Angry (1), Sad (2), Surprised (3), Neutral (4), Others (5).
  • The execution time for processing 100 × 100 pixel size is 10 ms on CPU.

Ada.py

An Improved Adaboost Algorithm inspired from Shen et al to select best N features from Gabor Filters Bank that distinguish between Face (affectNet) / No Face (CIFAR-10) after balancing both datasets.

Prerequisites

  • PyTorch
  • torchvision
  • NumPy
  • face_recognition package (For Detection (can be replaced with any detection Algorithm)

Run For a single Image

$ git clone https://github.com/marwankefah/emotionRecognition

$ python run.py --imgPath [path to image] --outPath [path to output]

Example 1

$ python run.py --imgPath face.jpg --outPath result.jpg

result

Example 2

$ python run.py --imgPath surprised.jpg --outPath result1.jpg

result1

Colab

Open In Colab
Make Sure GPU is enabled on COLAB to use CUDA-enabled GPU for processing

!pip install face_recognition

!git clone https://github.com/marwankefah/emotionRecognition

%cd emotionRecognition

!python run.py --imgPath [path to image] --outPath [path to output]