A modified technique for recognizing facial expression

Abstract

This paper consists of analysis of an algorithm dealing with facial expressions recognition. The algorithm has three major steps, initially image is processed, then the facial features are extracted and finally facial expression is recognized. In the initial processing stage the facial region is identified using a Haar cascade classifier. This facial region is passed on to the model trained by a CNN where facial features are matched with the features specified in the model. In the final step on the basis of comparison in the previous step the image is labelled and results are displayed. By the experiment results it is clear that the method specified in the paper can detect facial expressions very well.

Keywords

• facial expression recognition
• haar cascade
• cnn
• opencv