TY - GEN
T1 - Learning to Predict Risky Driving Behaviors for Autonomous Driving
AU - Hsu, Chih Chung
AU - Tseng, Wen Hai
AU - Yang, Hao Ting
N1 - Funding Information:
This study was supported in part by the Ministry of Science and Technology, Taiwan, under Grants MOST 108-2634-F-007-009, 107-2218-E-020-002-MY3, and 108-2218-E-006 -052.
Publisher Copyright:
© 2020 IEEE.
PY - 2020/9/28
Y1 - 2020/9/28
N2 - The most critical issue in the autonomous car is safety. Many kinds of research were proposed in recent years, such as car accident, obstacle, lane detection, and sign recognition, to study this issue. However, we can observe that some clues can be seen before the crash occurs. Several large-scale datasets were established by different research groups in recent years to study the driving behaviors to obtain better driving experience for the autonomous car. However, no dataset focuses on risky driving behaviors. risky and dangerous driving behavior will directly lead to car accidents. Once we can discover the risky driving behaviors in advance, it is possible to make more response time. In this paper, we collect 400 our own videos with car accidents and carefully annotate the dangerous behaviors, car accidents, and object contextual information for each video. We also investigate the preliminary approach to discover the cues of the common risky behaviors from the collected dataset. The initial experiments also show that the common dangerous behaviors mining can effectively increase the response time before the car accident occurs.
AB - The most critical issue in the autonomous car is safety. Many kinds of research were proposed in recent years, such as car accident, obstacle, lane detection, and sign recognition, to study this issue. However, we can observe that some clues can be seen before the crash occurs. Several large-scale datasets were established by different research groups in recent years to study the driving behaviors to obtain better driving experience for the autonomous car. However, no dataset focuses on risky driving behaviors. risky and dangerous driving behavior will directly lead to car accidents. Once we can discover the risky driving behaviors in advance, it is possible to make more response time. In this paper, we collect 400 our own videos with car accidents and carefully annotate the dangerous behaviors, car accidents, and object contextual information for each video. We also investigate the preliminary approach to discover the cues of the common risky behaviors from the collected dataset. The initial experiments also show that the common dangerous behaviors mining can effectively increase the response time before the car accident occurs.
UR - http://www.scopus.com/inward/record.url?scp=85098458712&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85098458712&partnerID=8YFLogxK
U2 - 10.1109/ICCE-Taiwan49838.2020.9258163
DO - 10.1109/ICCE-Taiwan49838.2020.9258163
M3 - Conference contribution
AN - SCOPUS:85098458712
T3 - 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020
BT - 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 7th IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020
Y2 - 28 September 2020 through 30 September 2020
ER -