Pain face estimation system for dementia patients based on video analysis
| Project/Area Number |
17K00442
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Web informatics, Service informatics
|
|---|
| Research Institution | Kochi University of Technology |
|---|
Principal Investigator |
KURIHARA Toru 高知工科大学, 情報学群, 准教授 (50401245)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
河野 崇 高知大学, 教育研究部医療学系臨床医学部門, 准教授 (40380076)
|
|---|
| Project Period (FY) |
2017-04-01 – 2020-03-31
|
| Project Status |
Completed (Fiscal Year 2019)
|
| Budget Amount *help |
¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000)
Fiscal Year 2019: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2018: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
Fiscal Year 2017: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
|
|---|
| Keywords | 痛み / 表情 / アテンション / LSTM / 認知症 / 画像 |
|---|
| Outline of Final Research Achievements |
Based on an attention mechanism that assigns weights to regions of images that are important for image description text generation, we have developed a new locally spatial attention mechanism for pain face estimation．We then learned the locally spatial attention regions that are important for end-to-end pain estimation.Pain estimation was performed by weighting the important areas of the face that are likely to produce expressions of pain. Furthermore, facial expressions are dynamic transformations of the face in the time domain, and the proposed network architecture incorporates a long-term short-term storage network (LSTM). As a result, it detects more fine-grained changes in the face region than conventional attention mechanisms, and it is possible to detect changes in the face area. We were able to improve the accuracy of frame-by-frame pain intensity estimation.
|
| Academic Significance and Societal Importance of the Research Achievements |
エンドツーエンドで顔表情からの痛みレベル推定のために重要な局所的空間的アテンション領域を学習し、顔の中でも痛みの表情が出やすい重要な領域に重みをつけ痛み推定を行うネットワーク構造を考案した。
このようなアテンション機構は、痛み以外の基本6表情を推定することにも用いることが可能であり、痛み顔に限らず表情認識の推定精度の向上に貢献しうるものである。
|
Report
(4 results)
Research Products
(5 results)
