As the need for an intelligent transport system is growing rapidly, vehicle detection has gained a lot of attention recently. This work presents an approach to improve the performance of YOLOv3 model architecture for detecting vehicle objects. The improvement is focused on detection speed and accuracy. The YOLOv3's backbone is changed by adopting MobileNets architecture. Its anchor boxes are also reselected so that they focus on detecting vehicle objects. Additionally, some additional post-processing methods are employed to validate the detected bounding box. Extensive experiments and analysis were carried out and they validated that it is promising to employ the proposed approach in a vehicle detection system.
Improving Performance of YOLOv3 for Vehicle Detection
2019-09-01
719693 byte
Conference paper
Electronic Resource
English
Vehicle detection method based on improved YOLOv3
IEEE | 2022
|Vehicle Detection Based on Improved Yolov3 Algorithm
IEEE | 2020
|Vehicle Detection in Aerial Images Based on YOLOv3
Springer Verlag | 2021
|