Vehicle-camel collision avoidance system in Libyan desert roads using computer vision technique

Travelling across the desert roads in Libya is faced with a major challenge represented by the spread of camels along these roads. Stray camels crossing these roads have caused countless accidents, many of them were unfortunately reported as fatal. Conventional solutions such as road signs, wildlife warnings and fencing the highway sides are either ineffective, expensive or inadequate. In this paper, a simple and low-cost automated system is proposed to detect the presence of an animal on the road to alert the vehicle's driver and hence preventing the vehicle-animal collision. Using image processing and computer vision techniques, the suggested system provides a method for determining the distance of the animal from the vehicle using a camera setup. More specifically, the Histogram of Oriented Gradients (HOG) scheme is used in this work as a feature descriptor for the purpose of object (animal) detection. Through Matlab simulation, the system is trained via positive and negative images of camels on highways. To validate the results even more, the proposed system is examined in various weather conditions in the targeted desert roads of Libya and the results have shown that it was able to distinguish the presence of animals with high accuracy.