Distracted Driver Behavior Detection Based-on An Improved YOLOX Framework
- Resource Type
- Conference
- Authors
- Wei, Yajuan; Guo, Zhaoli; Dai, Chuan; Chen, Minsi; Xu, Zhijie; Liu, Ying; Fan, Jiulun
- Source
- 2022 27th International Conference on Automation and Computing (ICAC) Automation and Computing (ICAC), 2022 27th International Conference on. :1-6 Sep, 2022
- Subject
- Components, Circuits, Devices and Systems
Computing and Processing
General Topics for Engineers
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Transportation
Training
Deep learning
Computational modeling
Roads
Benchmark testing
Performance gain
Data science
YOLOX
CBAM module
Distracted driving behavior
mAP
- Language
With the surge of the number of cars, road traffic accidents occur frequently because of drivers’ distracted attention and abnormal behaviors, which causes huge losses to people’s lives and property. To alleviate this issue, an improved deep learning algorithm based on YOLOX framework was proposed in this research to detect driving behavior changes in live. An attention mechanism – Convolutional Block Attention Module (CBAM) - was introduced in multiple scales of feature layers to form the backbone of YOLOX network. A widely used data science competition platform was adopted for distracted behavior model training. The State Farm Distracted Driver Detection Dataset was used for model validation and performance benchmarking. Experimental results have indicated promising performance gain using the devised model over the original YOLOX framework in terms of mAP and inference time.