Lin, CHUN PU
National of ChangHua University of Educcation Student
E-mail:lincfgate@gmail.com
WU, TUNGKUANG
National of ChangHua University of Educcation Professor
E-mail:tkwu@im.ncue.edu.tw
Abstract
In recent years, heavy motorcycles with engine displacement of 250cc or higher have been allowed on the roads in Taiwan. However, according to traffic regulations, these heavy motorcycles must be parked in designated car parking spaces. There is no clear regulation regarding the placement of the license plate for motorcycles with engine displacement of 250cc or higher. This has caused a problem for motorcycles to enter car parking facilities that use license plate recognition systems, resulting in competition for parking spaces between cars and motorcycles when there is a shortage of parking spaces on the roadside. This study proposes a solution to this problem by using rear license plate recognition for motorcycles and cars in parking facilities. However, recognizing license plates from the rear is more complicated than from the front, as the license plate may be affected by surrounding text and graphics. Therefore, this study proposes a rear license plate detection method to address this issue.
Keywords :Rear License Plate Detection, Differences Between Car and Motorcycle License Plates, Neural Networks, Transfer Learning, Fine-Tuning