| Co-Investigator(Kenkyū-buntansha) |
目加田 慶人 中京大学, 工学部, 教授 (00282377)
井手 一郎 名古屋大学, 数理・データ科学教育研究センター, 教授 (10332157)
平山 高嗣 人間環境大学, 人間環境学部, 教授 (10423021)
出口 大輔 名古屋大学, 情報学研究科, 准教授 (20437081)
川西 康友 国立研究開発法人理化学研究所, 情報統合本部, チームリーダー (50755147)
|
|---|
| Budget Amount *help |
¥44,330,000 (Direct Cost: ¥34,100,000、Indirect Cost: ¥10,230,000)
Fiscal Year 2021: ¥8,450,000 (Direct Cost: ¥6,500,000、Indirect Cost: ¥1,950,000)
Fiscal Year 2020: ¥9,880,000 (Direct Cost: ¥7,600,000、Indirect Cost: ¥2,280,000)
Fiscal Year 2019: ¥9,750,000 (Direct Cost: ¥7,500,000、Indirect Cost: ¥2,250,000)
Fiscal Year 2018: ¥9,360,000 (Direct Cost: ¥7,200,000、Indirect Cost: ¥2,160,000)
Fiscal Year 2017: ¥6,890,000 (Direct Cost: ¥5,300,000、Indirect Cost: ¥1,590,000)
|
|---|
| Outline of Final Research Achievements |
In-vehicle camera images and surveillance camera images can be of very low quality depending on the environment. We have developed a method to recognize these very low-quality images with high accuracy. Specifically, we developed (1) a spatial-temporal adaptive learning method that accumulates a vast amount of historical information and spatially distant information and utilizes such information, (2) a spatial-temporal fusion recognition process that improves recognition accuracy by integrating various spatial-temporal information, and (3) human assistance methods using recognition results that take into account human weaknesses such as ease of overlooking. We demonstrated the effectiveness of this method through experiments.
|
