Vehicle identification and distance estimation are perhaps some of the core problems in the Advanced Driver Assistance System (ADAS) developments worldwide. The paramount importance of the autonomous vehicle is safe driving in traffic, which is driven by fundamentals such as object identification and distance estimation to guarantee the protected separation between the vehicles and to ensure a safe ride on the roads. In view of the above problem, we propose a secure real-time vehicle tracking and distance estimation model for vehicles in motion in the light of Stereovision. The proposed model uses a 3D(stereoscopic) framework to capture the traffic participants and afterward recognize moving vehicles with the You Only Look Once (YOLO) V3 method. Subsequently, continuous video frame run-through calculation is done using optics-based estimation. Thus, we can determine the distance between the nearest car and object. YOLO is one of the best performing models for object discovery in a highly dynamic environment. In our case, it is used to locate the closest vehicle identification based on the location accuracy. YOLO processes every picture separately, even in a continuous video or frames. Because of its intricate inherent characteristics, even an identification of minor items will not be missed out. Thus, in this project, object recognition and distance estimation between two dynamic vehicles are carried out successfully.
Enabling Object Detection and Distance Calculation in AI based Autonomous Driving System
2022-10-16
1166066 byte
Conference paper
Electronic Resource
English
CNN-Based Object Detection and Distance Prediction for Autonomous Driving Using Stereo Images
Springer Verlag | 2023
|Autonomous Driving Object Detection Platform
Springer Verlag | 2024
|European Patent Office | 2017
|Compact radar sensors enabling autonomous driving
Springer Verlag | 2017
|