KEYWORDS:
ABSTRACT:
The goal of our study is placed at part of the project
Active Human Interface. We have set of 60 photos with faces expressing
6 basic emotions (happiness, surprise, anger, disgust, sadness, fear).
From these photos we obtain 30 x- and y- coordinates of facial characteristic
points representing 3 face components (eyes, eyebrows, mouth). After data
pre-processing we try to recognize emotions by neural network. We use fast-back
propagation algorithm for supervised learning. We test many different network
topologies. We reach correct recognition ratio about 75% and find many
difficulties concerning conveying human emotions. As well we use Kohonen's
self-organizing maps for unsupervised learning.