COMPARAITIVE ANATYSIS OF FACIAL EXPRESSION RECOGNITION USING HMM AND SVM

In this paper, we compare and investigate facial expression recognition system (FERS) based on well-known features using Hidden Markov Model (HMM) and Support Vector Machine (SVM) and also with combination of HMM and SVM. In recent years, there has been increasing usage of deep learning techniques in FERS however which also suffers from the problem of generalization. In this aspect this paper systematically investigates the performance of FERS using conventional AI techniques. We exploit well-known features landmark and texture based LBP. We also propose logical partitioning of face and obtained encouraging results. We have tested on dataset CK+ and spontaneous own created dataset named ‘OAK'. Experiments are conducted with different combinations of parameters to verify the efficiency of FER. This paper aims to compare the FERS performance using SVM and HMM. Results clearly reflect the efficiency of SVM with landmark feature.