Convolutional Neural Network to Modify the Restoration of a CCTV E-Ticket Image

Traffic violations are now increasingly worrying the local community. There are many methods that can be used to minimize this incident, one of which is the government creating an E-Ticket program with CCTV to detect the number plates of vehicles that violate traffic. However, the resulting images from CCTV can cause difficulties for the authorities, and this is because the resolution of images produced from CCTV is not optimal; therefore, in this research, a program was created that uses the Convolutional Neural Network method with SwinIR and uses a Transformer. The dataset used is from ATCS Bandung. The data is in the form of a screenshot photo. The aim is to increase the resolution of the images taken from the CCTV. The final result of image restoration was 400%, and the percentage for recognizing police number plates was 90%. The percentage from the amount of clearly visible data/number of datasets x 100%.