Application of Fatigue Driving Detection Based on Edge Computing

Driving fatigue causes many traffic accidents. It’s crucial to detect driver fatigue to ensure traffic safety quickly. Existing detection schemes depend on cloud computing, which has issues like high latency, network instability, and privacy concerns. This paper proposes an edge computing scheme based on HUAWEI Atlas200DK. It combines the CANN framework and DVPP image processing unit to achieve local fast inference and avoid data upload issues. Experiments show this scheme’s response latency is 4-5 times faster than the cloud-based GPU scheme. Edge computing’s efficacy in fatigue driving detection is clear, offering high real-time performance, low power, stability, and privacy protection.