MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641112614 A) filed by Vallurupalli Nageswara Rao Vignana Jyothi Institute Of Engineering And Technology on September 19, 2026, for A Generative Adversarial Network-Based Intelligent License Plate Recognition System Using Yolov8 Detection And Ocr-Based Text Extraction.
Inventor includes Dr. Harshavardhan Awari.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: ABSTRACT [0016] The present invention provides a modular automatic number plate recognition system that addresses limitations arising from scarce training data, diverse plate designs, and adverse environmental conditions such as poor lighting, motion blur, and occlusions. The framework employs generative adversarial networks to augment the training dataset with synthetic license plate images spanning varied styles, lighting, and backgrounds, while ESRGAN enhances low-resolution or noisy video frames prior to detection. YOLOv8 performs real-time detection of vehicles and license plates on the enhanced frames, after which EasyOCR extracts characters from the cropped plate regions with resilience to font variations, distortions, and partial occlusions. An interpolation module restores continuity for plates that are momentarily missed across consecutive frames. Detection results are overlaid on the original video and exported as structured CSV logs. Experimental evaluation demonstrates that the integrated pipeline achieves 97.6 percent accuracy, 98.4 percent precision, and 94.2 percent recall, surpassing comparative methods that rely on earlier detectors or conventional OCR. The system operates efficiently on moderate hardware without extensive manual preprocessing, rendering it suitable for intelligent traffic monitoring, automated toll collection, parking management, and related smart transportation applications.
Disclaimer: Curated by HT Syndication.
