Facial expression recognition using image processing techniquesand convolutional neural networks

In this thesis, image processing techniques and machine learning techniques have been examined to detect facial emotions. The studies carried out to obtain the feature vectors and determine the emotion expressions on the face are given in stages. In the applications and categories, CK + image set consisting of 981 images containing human faces in 7 emotion categories was used. The feature data set was created from the CK + data set by using HOG, LBP, Geometric Features and Wavelet Transform methods. The feature data set is classified with Support Vector Machines, K-Nearest Neighborhood Algorithm and Logistic Regression classification methods, and face expression detection success cases are given comparatively. The original state of the images in the data set and the success rates in cases where HOG, LBP and Wavelet transform feature images were used were compared. The success of the convolutional neural networks (CNN) method in detecting emotion expressions is discussed.