A Review on Road Accident Detection: Learning Techniques and IoT based Techniques

Road accidents are the leading cause of fatalities and injuries worldwide. Due to the tremendous growth in road accidents, road safety gains significant attention from industries and researchers. More specifically, the valuable lives are lost due to the delay in emergency department response and the lack of information sharing about the location and severity of an accident. To solve this issue, extensive review has been conducted on the techniques available on traffic accident detection and information communication systems. Numerous effective methods have been developed to reduce road accidents and save injured lives by IoT and learning-based techniques. Various limitations occur while processing these techniques such as low accuracy rate with signal transmission blockage, high false alarm rate, communication difficulties due to damaged sensors, and so on. Hence, this research presents an intensive review focusing on road accident detection and accident warning systems. Further, this research offers a critical review of several techniques available for detecting road accidents by highlighting their pros, cons, and problems that must be resolved to guarantee traffic safety and save lives. In this review, the detection methods are analyzed with different datasets used by deep learning classifiers, machine learning classifiers, and IoT-based detection systems. Throughout the entire review, the detection methods are analyzed which aids future researchers to implement efficient road detection models in every vehicle and thus provides an accident-free environment. Moreover, the insights obtained from this review assist in improving the systems for detecting road accidents, resulting in improved public traffic safety.