TY - GEN
T1 - Top-Down Driver Head Orientation Detection Method Based on Monocular Camera and Virtual Scene Generation
AU - Shi, Jiangnan
AU - Tang, Jingyu
AU - Hu, Menghan
N1 - Publisher Copyright:
© 2022, Springer Nature Singapore Pte Ltd.
PY - 2022
Y1 - 2022
N2 - Head posture assessment plays a very important role in the fields of driver attention detection, driving behavior analysis, and driving habit analysis. The mainstream head posture assessment methods use cameras to capture all the driver’s front face videos, this approach has a natural drawback that it requires a lot of facially sensitive data which is contrary to the increasing awareness of privacy protection. Considering privacy protection, we innovatively apply a monocular camera to acquire an image of the driver’s head with top-down shooting view and develop a corresponding TOP OF HEAD ONLY (TOHO) algorithm to estimate the head orientation. Subsequently, the virtual scene generation technique is leveraged to generate diverse model data. The generated data is afterwards used to validate the effectiveness of the proposed TOHO algorithm. The experimental results show that the error of the head attention angle calculated by the proposed algorithm is within 15∘ compared to the labeled data.
AB - Head posture assessment plays a very important role in the fields of driver attention detection, driving behavior analysis, and driving habit analysis. The mainstream head posture assessment methods use cameras to capture all the driver’s front face videos, this approach has a natural drawback that it requires a lot of facially sensitive data which is contrary to the increasing awareness of privacy protection. Considering privacy protection, we innovatively apply a monocular camera to acquire an image of the driver’s head with top-down shooting view and develop a corresponding TOP OF HEAD ONLY (TOHO) algorithm to estimate the head orientation. Subsequently, the virtual scene generation technique is leveraged to generate diverse model data. The generated data is afterwards used to validate the effectiveness of the proposed TOHO algorithm. The experimental results show that the error of the head attention angle calculated by the proposed algorithm is within 15∘ compared to the labeled data.
KW - Driver attention detection
KW - Head pose assessment
KW - Image processing
KW - Privacy preservation
KW - Virtual scene generation
UR - https://www.scopus.com/pages/publications/85128894007
U2 - 10.1007/978-981-19-2266-4_27
DO - 10.1007/978-981-19-2266-4_27
M3 - 会议稿件
AN - SCOPUS:85128894007
SN - 9789811922657
T3 - Communications in Computer and Information Science
SP - 349
EP - 360
BT - Digital TV and Wireless Multimedia Communications - 18th International Forum, IFTC 2021, Revised Selected Papers
A2 - Zhai, Guangtao
A2 - Zhou, Jun
A2 - Yang, Hua
A2 - An, Ping
A2 - Yang, Xiaokang
PB - Springer Science and Business Media Deutschland GmbH
T2 - 18th International Forum of Digital Multimedia Communication, IFTC 2021
Y2 - 3 December 2021 through 4 December 2021
ER -