Text Recognition on Curved Metal Surface: An Enhanced Approach using EasyOCR and OpenCV

Text is a fundamental system of symbolic representation, and it plays a major role in human culture and communication. Its ability to convey the rich semantic information has become very essential in vision-based applications. With the advancement of deep learning, computer vision techniques particularly in text detection and recognition has undergone significant evolution. This paper investigates the transformative impact of deep learning on scene text detection and recognition, specifically on curved metal surfaces. By leveraging Easy Optical Character Recognition (EasyOCR) and OpenCV, the proposed approach achieves an impressive accuracy of 90% for text recognition in challenging curved surfaces. Additionally, this study explores to optimization techniques designed to enhance detection precision and examines the scalability of deep learning models in real-world applications.