Number plate detection through machine learning approach

An efficient Automatic Number Plate Recognition (ANPR) system is required in the futuristic world. Due to the fact that each county has its own number plate type and style, an unrestricted Number Plate Detection system is still unavailable. Because to the lack of data and the varied plate formations, there hasn't been much work done on Pakistani number plates. To address this problem, a Pakistani vehicle dataset has been collected with diverse plate designs and used it to construct an innovative ANPR system. Using machine learning technique, the suggested framework locates the number plate region. It employs a CNN model, employs strong preprocessing techniques on the recovered plate region, and then uses OCR (Optical Character Recognition) Python-Tesseract to recognize the plate label. The suggested ANPR framework, which uses the CNN and OCR Tesseract for detection and identification, has good accuracy and inference time for a wide range of lighting and style of Pakistani number plates, and may be utilized to construct a real-time system, according to the comparative analysis. Researchers developing ANPR for countries with similar demanding vehicle number plate formats and designs may benefit from the suggested ANPR framework.