Automatic license plate recognition (ALPR) is the extraction of vehicle license plate information from an image. The extracted information can be used in many applications such as electronics payment systems at toll plazas and parking areas. It can also be used for freeway and urban traffic surveillance. ALPR as a real life application has to quickly process license plates under different environmental conditions such as day, night, rainy etc. The quality of the image captured play a major role in correctly identifying the registration number. The number plates might use different fonts and colors as well. Also some license plates can be partially occluded by dirt, lighting, and towing accessories on the car. In this project, given an image of a license plate we aim to correctly identify the registration number using Image segmentation techniques and Convolutional Neural Network Models(CNN). Later we aim to improve the prediction accuracy by data augmentation, feature engineering and ensemble machine learning methods.

Our ALPR system can be split into two stages: • Character segmentation - extract the alphanumeric characters from the plate • Character recognition - recognize each individual character

Please go through the self-explained ALPR.ipynb for more details