In this paper, we propose an understanding presumption system from facial images using a three-layered neural network. It can presume a degree of understanding from facial expressions; it can recognize whether a person understands a question or not. Feature points are located on each facial image and are used to extract an expression information. The expression information is given as an input and the system presumes the degree of understanding based on the facial images into 5 levels, from NOT UNDERSTAND to WELL UNDERSTAND. The network is learned using Back Propagation algorithm. The average presumption rates of the proposed system was 71.3%.
|出版ステータス||Published - 1999|
|イベント||International Joint Conference on Neural Networks (IJCNN'99) - Washington, DC, USA|
継続期間: 1999 7月 10 → 1999 7月 16
|Other||International Joint Conference on Neural Networks (IJCNN'99)|
|City||Washington, DC, USA|
|Period||99/7/10 → 99/7/16|
ASJC Scopus subject areas