The research focuses on enhancing license plate recognition by using a multi-angle view model. This method extracts descriptive features of the text components in the license plate from multiple frames of distinct perspectives. The study introduces three viewpoints to identify neighboring components that aid in restoring text components from the same license plate line. The CnOCR method is then used for text recognition within license plates. The proposed method showed promising results in experiments conducted on a self-collected dataset and the Stanford Cars Dataset. The study contributes to the field of Intelligent Transportation Systems by improving the accuracy of license plate recognition.
Publication date: 22 Sep 2023
Project Page: https://arxiv.org/abs/2309.12972v1
Paper: https://arxiv.org/pdf/2309.12972
