A method of learning based boosting in multiple classifier for color facial expression identification

Abstract

An automatic color facial expression recognition system has been designed and developed using multiple classifier classifications. This facial expression recognition system involves extracting the most communicative facial parts such as forehead, eyes with eyebrows, nose and mouth. Then these extracted features are trained individually using different classification system. Finally, a super classifier fuses the conclusions drawn by individual classifier which results in a final decision. This improves the overall system performance significantly in terms of accuracy, precision, recall and F-score with holdout method. Experimental result shows about 98.75% accuracy. The learning as well as performance evaluation time of the system is affordable.