| Project/Area Number |
14580224
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Educational technology
|
|---|
| Research Institution | KYUSHU UNIVERSITY |
|---|
Principal Investigator |
SUGANUMA Akira Kyushu University, Department of Information Systems, Associate Professor, 大学院・システム情報科学研究院, 助教授 (70235852)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
TANIGUCHI Rin-ichiro Kyushu University, Department of Information Systems, Professor, 大学院・システム情報科学研究院, 教授 (20136550)
ARITA Daisaku Kyushu University, Department of Information Systems, Research Associate, 大学院・システム情報科学研究院, 助手 (70304756)
|
|---|
| Project Period (FY) |
2002 – 2004
|
| Project Status |
Completed (Fiscal Year 2004)
|
| Budget Amount *help |
¥3,000,000 (Direct Cost: ¥3,000,000)
Fiscal Year 2004: ¥800,000 (Direct Cost: ¥800,000)
Fiscal Year 2003: ¥900,000 (Direct Cost: ¥900,000)
Fiscal Year 2002: ¥1,300,000 (Direct Cost: ¥1,300,000)
|
|---|
| Keywords | Image processing / Distant lecture / Supporting system / Estimation of a lecture / Estimating teacher's behavior / Estimation student's behavior / Automatic camera control / 音処理応用 |
|---|
| Research Abstract |
We have developed a method estimating a teacher's behavior from video scene of a lecture. When a lecture scene is captured automatically, a capturing system has to control an active camera according to the teacher's behavior. So that our method has to estimate the teacher's behavior as accurately as possible. We have solved this problem adopting both image processing and sound processing.
We have developed a method more accurately extracting an object explained by the teacher based on both categorization of the teacher's behavior by image processing and contents of the teacher's utterance. We observed some lecture videos to investigate the teacher's behavior. We found out that the behavior of the teacher could be categorized into three categories for automatically capturing a lecture. Our method estimates an explained object based on categorized behavior, the location of the latest object written on a blackboard and the teacher's position. And we investigated a relation between pointing words in the teacher's utterance and relative position from him to the explained object. We also selected some useful words for estimating the explained object.
We have developed a method detecting a students attitude in a distant classroom where PCs set up for students. We designed our method on the assumption that each student assigns one PC which is equipped with a USB camera. Our system can detect browsed Web pages, an application running on the student's PC, and student's facial direction. Detecting the facial direction is based on the template matching method.
We have developed a real-time analysis method of complexion of a distant classroom. Designing our method, we assumed that almost all students study earnestly and tend to behave synchronously. Our method categorizes student's action such as browsing a material page explained before to estimate complexion of a classroom. Results of analyzing the complexion are sent to the teacher successively.
|
