深層学習と圧縮センシングを融合した革新的超低消費電力イメージングシステムの実現
| Project/Area Number |
22K12101
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 61010:Perceptual information processing-related
|
|---|
| Research Institution | Hosei University |
|---|
Principal Investigator |
周 金佳 法政大学, 理工学部, 准教授 (50723392)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
谷口 一徹 大阪大学, 大学院情報科学研究科, 准教授 (40551453)
|
|---|
| Project Period (FY) |
2022-04-01 – 2025-03-31
|
| Project Status |
Granted (Fiscal Year 2023)
|
| Budget Amount *help |
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2024: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2023: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Fiscal Year 2022: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
|
|---|
| Keywords | Image sensing / Deep learning / Compressive sensing |
|---|
| Outline of Research at the Start |
This research proposes a new optical coding with AI based measurement coding and smart sparse recovery system that can greatly reduce the sensing power and compression power at the same time. It is the first time to design a sensing pattern that can efficiently compress the signal during sensing.
|
| Outline of Annual Research Achievements |
The following tasks have been completed. (1) On the encoder side, building upon the adaptive sensing technique developed last year, we further optimized it and applied an edge detection-based and region-of-interest detection based adaptive sensing framework to enhance the quality of the reconstructed videos. (2) On the decoder side, we proposed an adaptive stage-activated unfolding network to adaptively control the complexity of reconstruction. (3) We optimized the entire system at the system level to realize a computer vision-oriented compressive sensing system. (4) To reduce the power consumption of the whole system, we proposed a framework for partial pre-calculation-based image encoding and decoding.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
As planned, the adaptive sampling algorithms based on key information extraction achieved excellent results in reducing computational complexity and improving the quality of reconstructed images/videos. Furthermore, the system-level optimization also yielded positive outcomes. We will continue to improve the performance of the whole system.
|
| Strategy for Future Research Activity |
This project is divided into three main tasks, each building upon the progress made in the previous stages. The first task of developing measurement coding system at the encoder side has already been successfully completed. In the second task, we focused on extracting moving objects, edge information and region-of-interest as key information during the FY2022 and FY2023. Our plan in FY2024 is to refine and enhance the key information extraction algorithms, incorporating new techniques to effectively selecting the sampling ratio while maintaining video quality. Additionally, we aim to optimize the entire system for reducing the power consumption.
|
Report
(2 results)
Research Products
(9 results)
