A Deep Learning-Based Car Accident Detection Framework Using Edge and Cloud Computing

Abstract

The ever-changing technology landscape has seen a significant breakthrough with the introduction of edge computing. This innovation has revolutionized various domains, and one of its critical applications is in the domain of accident detection. Edge computing can help enhance road safety and emergency response by enabling real-time processing and analysis of sensory information from onboard sensors, cameras, and other connected devices. By integrating edge computing into accident detection systems, we can overcome the limitations of conventional centralized cloud-based methods and create a safer transportation network. In this article, we have proposed an accident detection framework using Deep Learning (DL) in the edge cloud environment. For accident detection, we have used a Convolutional Neural Network (CNN)- based DL model. The DL model detects the accident in the edge node which is near the data source. The proposed framework provides low latency, minimal network usage, and lower execution time as compared to only cloud-based deployment. Additionally, the proposed accident detection model is accurate up to 95.91% with Precision 0.9574, Recall 0.9574 and F1 score 0.9574 in the cloud-edge environment.

Keywords

• Edge computing
• cloud computing
• deep learning
• accident detection
• road safety
• 2D CNN