Answers are grounded in this paper and cite their sources.
In this paper we propose an Automatic License Plate Recognition (ALPR) system for unsupervised parking lot applications. The main objective is to develop a system which is implementable on embedded devices, specifically a Raspberry-PI3. ALPR consists of two main stages: (I) Locating the plate and (II) Optical Character Recognition (OCR). Considering the recent growth and success of deep learning methods, especially convolutional neural networks (CNN), in our system, we used the Single Shot Detection (SSD) architecture along with the MobileNet feature extractor to detect the plate in the image captured by the camera and the LPRNet network for OCR. The proposed method is robust, accurate, computationally inexpensive and able to perform in Real-time. The system achieves 79.86% end-to-end accuracy on our dataset and successfully performs in real-time on a Raspberry-PI3. For training the OCR network, we generated and used 130k synthetic license plate images. We also introduce a dataset containing 1500 images with various camera zoom, lighting and viewpoint conditions.
No takes yet. Share an insight, caveat, or question.
Alborzi et al. (2019) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: