This document presents a survey of existing automatic number plate recognition (ANPR) systems and proposes a new approach using k-nearest neighbors (k-NN), OpenALPR, and convolutional neural networks (CNNs). It first discusses challenges with license plate recognition in India due to regional fonts. Existing ANPR systems are reviewed along with their drawbacks. The proposed method involves detecting license plates in images and videos, extracting characters, and recognizing plates using k-NN, OpenALPR and a CNN model. Results of the three approaches on test data will be analyzed to determine the most accurate one, considering factors like character recognition accuracy and processing time.