| Project/Area Number |
18K19818
|
|---|
| Research Category |
Grant-in-Aid for Challenging Research (Exploratory)
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Medium-sized Section 61:Human informatics and related fields
|
|---|
| Research Institution | Osaka University |
|---|
Principal Investigator |
Nagahara Hajime 大阪大学, データビリティフロンティア機構, 教授 (80362648)
|
|---|
| Project Period (FY) |
2018-06-29 – 2020-03-31
|
| Project Status |
Completed (Fiscal Year 2019)
|
| Budget Amount *help |
¥6,240,000 (Direct Cost: ¥4,800,000、Indirect Cost: ¥1,440,000)
Fiscal Year 2019: ¥2,730,000 (Direct Cost: ¥2,100,000、Indirect Cost: ¥630,000)
Fiscal Year 2018: ¥3,510,000 (Direct Cost: ¥2,700,000、Indirect Cost: ¥810,000)
|
|---|
| Keywords | コンピュテーショナルフォトグラフィ / ディープラーニング |
|---|
| Outline of Final Research Achievements |
Deep learning is getting popular in computer vision and it drastically improve the performance of object recognition, scene understanding and image reconstruction etc. However, regular deep learning is applied to the digital domain of a camera pipeline and ignored a physics layer such as optics and sensor of a camera. In this paper, we proposed a framework for modeling the camera pipeline including the physics layer as well as the digital layer and optimize the camera design parameters and task model by learning. We have demonstrated this framework to the compressive video sensing and action recognition tasks and show the effectiveness of the approach.
|
| Academic Significance and Societal Importance of the Research Achievements |
これまでのカメラの設計は，サンプリング理論やノウハウにより設計者の手動により設計されてきた．これに対して，本研究では，データ駆動におる学習アプローチにより，カメラの設計パラメータを最適化することにより，応用に即したハードウェアを設計することで性能向上を行った．このようなアプローチは，Deep sensingやDeep opticsなどと呼ばれ，後追い研究を呼び最近の研究の新しい流れのひとつとなっている．
|
