| Project/Area Number |
26280057
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Partial Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Perceptual information processing
|
|---|
| Research Institution | Nagoya University |
|---|
Principal Investigator |
Kato Jien 名古屋大学, 情報科学研究科, 准教授 (70251882)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
井手 一郎 名古屋大学, 情報科学研究科, 准教授 (10332157)
|
|---|
| Co-Investigator(Renkei-kenkyūsha) |
MASE KENJI 名古屋大学, 大学院情報科学研究科, 教授 (30345855)
|
|---|
| Project Period (FY) |
2014-04-01 – 2017-03-31
|
| Project Status |
Completed (Fiscal Year 2016)
|
| Budget Amount *help |
¥16,510,000 (Direct Cost: ¥12,700,000、Indirect Cost: ¥3,810,000)
Fiscal Year 2016: ¥3,250,000 (Direct Cost: ¥2,500,000、Indirect Cost: ¥750,000)
Fiscal Year 2015: ¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000)
Fiscal Year 2014: ¥8,710,000 (Direct Cost: ¥6,700,000、Indirect Cost: ¥2,010,000)
|
|---|
| Keywords | 人体パーツに注目した歩行者検出 / 危険予測のためのビデオデータベースの作成 / 歩行者行動範囲の予測 / 事故リスクの見積もり / 単眼カメラによる地面の推定 / 安全運転支援 / 歩行者危険行為の早期検出 / 人体パーツに注目した歩行者の検出 / 危険予測のためのビデオデータベースの構築 / 事故リスクのレーティング / 画像、文章、音声等認識 / 高度道路交通システム（ITS） / 歩行者早期検出 / 歩行者行為の早期検出 / MLDP特徴 / 近似スパースコーディング / 歩行者詳細属性の識別 / 路面位置推定 / 単眼カメラ / ホモグラフィ / サーフェスレイアウト / 似スパースコーディング |
|---|
| Outline of Final Research Achievements |
In this work, we study the task of early detection of pedestrians’ dangerous behavior with the purpose of safe-driving assistance. We followed our plan and worked in the following topics: (1) early detection of pedestrian; (2) early detection of pedestrian's behavior; (3) dangerous level estimation with pedestrians’ multiple attributes. Related to (1), we developed occlusion-robust pedestrian detection approaches, and fast road surface estimation approach which can be used to speedup pedestrian detection algorithms. Related to (2), we collected a large-scale data set for driving dangerous estimation, and developed an attribute estimation approach which can be used to predict pedestrian behavior and an attainable region estimation approach which can estimate pedestrian’ area of behavior on the road. (3) we developed an end-to-end learning approaches for estimating various driving dangerous, including pedestrian’s danger behaviors, based on the data set we collected in.
|
