Expression Recognition
Gabor filters applied to a sample BSAF image.
Classified which of 6 expressions (Joy, Sorrow, Disgust, Anger, Surprise, Fear) were being displayed in grayscale images of subjects’ faces.
We trained one-vs-many linear SVMs on Gabor filters applied to the input images. We used existing standard datasets (i.e., Cohn-Kanade and JAFFE) as well as a self-collected 47-subject dataset: the Berkeley Students and Friends (BSAF).
