Last Decade in Vehicle Detection and Classification: A Comprehensive Survey

Due to the ever increasing traffic on roads, there has been a pressing need for Automatic Vehicle Detection (AVD) systems, so that the real-time traffic can be observed as well as managed in an efficient way. To address this research problem, several techniques have been put forward in the last decade by researchers across the world. In this article, a comprehensive survey has been introduced comprising various methods used for AVD and classification during the last 10 years. Besides, these methods are categorized following the two learning eras: Machine learning and Deep learning. We have also presented a comparative study of the different methods and discussed their pros and cons. Various datasets, proposed and used over the last 10 years, are also reported in this survey. Although there are a few survey papers published on AVD, this survey paper offers some deeper insights into the methods proposed by the researchers in the last decade. Finally, the article has been concluded with critical analysis of the methods and some important future research directions toward AVD. Researchers who want to either use or construct dependable AVD systems that fulfil their needs might use the results of this study as a guide.

Automated summary

This article introduces various methods used for AVD and classification over the past 10 years, compares different methods, discusses their pros and cons, and proposes future research directions towards AVD.