With the continuous growth of car ownership, intelligent transportation has penetrated into our daily life. As an important part of intelligent transportation, car detection has also been developed rapidly. It plays a vital role on the planning of urban public transportation and brings great convenience for citizens to commute. Due to the extremely complex urban conditions, car detection encounters many difficulties. By analyzing a large amount of vehicle color difference data, it draws the conclusion that illumination is the main factor of affecting detection and recognition. Based on the convolutional neural network framework, this paper focuses on low-light enhancement and car recognition, aiming to realize the task of car recognition in more complex low-light situations. Notably, in those scenarios, the YOLOv4 model with a basic training set can recognize automobiles well.
Vehicle detection in dark weather based on YOLO and MSRCR method
Second International Conference on Optics and Image Processing (ICOIP 2022) ; 2022 ; Taian,China
Proc. SPIE ; 12328
10.09.2022
Aufsatz (Konferenz)
Elektronische Ressource
Englisch