講演抄録/キーワード |
講演名 |
2022-02-28 15:20
Responsibility Assessment in Crossroad Accident Using Object Recognition and Knowledge System ○Yawovi Agbewonou Helton・Kikuchi Masato・Ozono Tadachika(NITech) AI2021-22 |
抄録 |
(和) |
Car crashes are increasing year by year, and they represent the side effect of the increasing number of vehicles in the modern world. Usually, vehicle collisions lead to injuries and deaths, with a devastating impact on victims and their relatives. After a crash, the police have to investigate to know the circumstances of the incident and determine each actor’s responsibility. The police need support systems to conduct these investigations more efficiently. In this work, we realized a system that helps evaluate each actor’s responsibility in a crossroad crash. We proposed a heuristic approach that can evaluate actors’ responsibilities when a crash occurs, thanks to the usage of the driving recorder video of the crash as the data source. The system uses the crash video as the input data source and outputs the evaluation of each actor’s responsibility in the crash. It consists of three agents: (1) Crash time detection and crash video split into images; (2) Traffic sign detection in the crash video; (3) Responsibility evaluation using a knowledge system. In this paper, we describe our system, evaluate it through experiments, and list our current improvement tasks. |
(英) |
Car crashes are increasing year by year, and they represent the side effect of the increasing number of vehicles in the modern world. Usually, vehicle collisions lead to injuries and deaths, with a devastating impact on victims and their relatives. After a crash, the police have to investigate to know the circumstances of the incident and determine each actor’s responsibility. The police need support systems to conduct these investigations more efficiently. In this work, we realized a system that helps evaluate each actor’s responsibility in a crossroad crash. We proposed a heuristic approach that can evaluate actors’ responsibilities when a crash occurs, thanks to the usage of the driving recorder video of the crash as the data source. The system uses the crash video as the input data source and outputs the evaluation of each actor’s responsibility in the crash. It consists of three agents: (1) Crash time detection and crash video split into images; (2) Traffic sign detection in the crash video; (3) Responsibility evaluation using a knowledge system. In this paper, we describe our system, evaluate it through experiments, and list our current improvement tasks. |
キーワード |
(和) |
Crossroad car crash / Knowledge system / Machine learning / Object detection / Responsibility evaluation / / / |
(英) |
Crossroad car crash / Knowledge system / Machine learning / Object detection / Responsibility evaluation / / / |
文献情報 |
信学技報, vol. 121, no. 382, AI2021-22, pp. 59-64, 2022年2月. |
資料番号 |
AI2021-22 |
発行日 |
2022-02-21 (AI) |
ISSN |
Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
PDFダウンロード |
AI2021-22 |
|