Statistical Machine Learning: Facial Emotion Recognition and Classification

Date:

  • Constructed facial emotions extracted from 2,000+ training images with 20+ different emotions, then reduced the dimension from 6,006 to 50 by PCA and other creative feature selection methods
  • Achieved 70+% accuracy and high time efficiency with <10s for 500 images dataset by constructing an innovative combination classifier of LDA and SVM to improve special emotions classification