Traffic Object Detection and Distance Estimation Using YOLOv3 2022-28-0120

According to the National Highway Traffic Safety Administration (NHTSA), many people are killed each year in automobile collisions, this may be because of distractions like inattention, sleepiness, drowsiness falls, talking or texting using mobile phones, eating and drinking, talking to passengers in the vehicle, adjusting the radio, etc… Improvements in Advanced driving and correspondence innovations have given broad and propelled changes. The use of new innovations gives incredible advantages to people, organizations, and governments. This paper works on Object detection to assist drivers and autonomous vehicles in estimating crash risks using dash cameras on vehicles and Machine learning algorithms. Although these technologies (Object detection) are increasingly available, the cost-effective method still remains a challenge. I propose an approach to understand the situation with Object detection using YOLOv3 algorithm that uses Draknet-53 CNN to detect and locate the objects like pedestrians, animals, vehicles, etc. All identified information can provide awareness to drivers and autonomous vehicles for identifying crash risks from the traffic. But driver looking into display screen is dangerous, so this paper work has added advantage of warning the driver if the object detected is too close to collide, distance measurement is required for ADAS (advanced driver assistance systems) to warn the driver and avoid collisions and tailgating detection. Generally, Radars and LIDARs are mainly used for this, which are expensive. There are many methods to estimate the distance such as ultrasonic ranging, laser ranging, etc. This paper uses Object image and focal distance relationship to measure the distance.