Analysis of Statistical and Artificial Intelligence Algorithms for Real-Time Speed Estimation Based on Vehicle Detection with YOLO

Automobiles have increased urban mobility, but traffic accidents also increased. Therefore, road safety is a significant concern involving academics and government. Transit studies are the main supply for studying accidents, congestion, flow traffic, allowing understanding of flow. They require special equipment (sensors) to measure car’s speed. With technological advances, artificial intelligence, videos, it possible estimate speed in real-time without modifying installed infrastructure. We need employ public databases that provide reliable monocular videos generate automated studies. The problem estimation with camera involves synchronizing data recording, tracking, detecting vehicles over considering lanes distance between cars. Usually, set constraints considered, such as calibration, flat roads, including methods based on homography augmented intrusion lines, patterns or regions, prior knowledge about actual dimensions some objects. In this paper, we present system generates dataset from recorded highway—obtaining 532 samples; separated vehicle’s detection by lane, estimating its use compare five different statistical three machine learning evaluate their accuracy cars’ real-time. Our vehicle requires feature extraction process using YOLOv3 Kalman filter detect track vehicles. Linear Regression Model (LRM) yielded best results obtaining Mean Absolute Error (MAE) 1.694 km/h center lane 0.956 last lane. were compared several state-of-the-art works, having competitive performance. Hence, LRM fast real time does not high computational resources future hardware implementation.