Recognition of facial expressions using associative memory

We extract the movements of 8/spl times/10 regions of the face by using optical flow and simplify the information by considering whether a particular region of the face moved or not. By applying this information in a Hopfield neural network based on a discrete model, we try to absorb the differences between individuals and the degree of feelings effectively. We carried out an experiment with 144 examples of image data. The results of before and after applying the data in the Hopfield neural network were 61.8% and 71.5% respectively.

[1]  K. Shirai,et al.  Extraction of speaker's feeling using facial image and speech , 1995, Proceedings 4th IEEE International Workshop on Robot and Human Communication.

[2]  J J Hopfield,et al.  Neurons with graded response have collective computational properties like those of two-state neurons. , 1984, Proceedings of the National Academy of Sciences of the United States of America.

[3]  J. Hopfield.  Neurons withgraded response havecollective computational properties likethoseoftwo-state neurons , 1984 .

[4]  Berthold K. P. Horn,et al.  Determining Optical Flow , 1981, Other Conferences.

[5]  J J Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities. , 1982, Proceedings of the National Academy of Sciences of the United States of America.