Vehicle detection approaches have attracted great attention of the autonomous systems community in recent years. Related research in vehicle detection, line detection and driver intention recognition have yielded various solutions for semi-autonomous and fully autonomous vehicles. The main motivations of the researches in this field are to lessen the loss of human lives and property damages as a result of vehicle accidents, to reduce the role of the drivers in transportation, and to optimize the time and energy consumptions in dynamic environments. To achieve such targets, regional and international organizations like the European Union regulate automotive manufacturers to standardize advanced autonomous transportation systems equipped with the camera to sense the dynamic environments. Such regulations will form the basis of the autonomous vehicles in the coming years. In this paper, Fast Regions for Convolutional Neural Network (FR-CNN) based vehicle detection algorithm is applied to detect the neighbor vehicle, the lanes and to recognize the driver intentions. After the vehicle detection, the movements of the vehicles in the traffic are tracked and supported to instant intention of the neighbor vehicle is assessed to regulate the motion of the vehicle accordingly. The output of the study, it is aimed to minimize the damage by warning the user of a possible accident and that autonomous technologies have a significant importance in these applications.
Anahtar Kelimeler: Camera vision, Driver intention, Fast R-CNN, Lane detection, Vehicle detection, Vision-based detection
|
