2019 Fiscal Year Final Research Report
Development of animal tracking system in real situation
| Project/Area Number |
17K00240
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Perceptual information processing
|
|---|
| Research Institution | University of Miyazaki |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
坂本 信介 宮崎大学, 農学部, 准教授 (80611368)
|
|---|
| Project Period (FY) |
2017-04-01 – 2020-03-31
|
| Keywords | 小動物追跡 / エネルギー最小化原理 / スパース最適化 / SCMA法 / 複雑背景 / 深層学習 |
|---|
| Outline of Final Research Achievements |
In this research, I develop and evaluate a new discrimination method, called as Sparse Collaborative Mean Attraction Method (SCMA). The SCMA method shows good performance for discrimination especially when the number of training data per category is very small. I also apply the SCMA method for tracking animals in a cage. It shows better performance than previous methods, such as particle filter or AKAZE feature tracking. I am now implementing the developed method on smart phone, which can encourage people to use the developed method in real situation.
|
| Free Research Field |
画像認識
|
| Academic Significance and Societal Importance of the Research Achievements |
エネルギー最小化の枠組みにスパース最適化を導入し，種々の制約条件を一つの枠組み内で表現した上で，その中の有効な条件を選択的に最適化して解を求める Sparse Collaborative Mean Attraction法(SCMA法)を提案した．SCMA法は，特に学習データが少数の場合に汎用的に有効な識別器であることを実験により示した．また，SCMA法を実際に動物追跡に適用し，従来手法よりも正しく追跡が行えることを実験により示した．さらに，開発したシステムは，スマートフォン上への実装を進めており，実装が完了すれば研究成果を広く利用できる形で公開できる．
|
