计算机科学
解码方法
目标检测
人工智能
计算机视觉
编码(集合论)
多线程
模式识别(心理学)
算法
线程(计算)
操作系统
集合(抽象数据类型)
程序设计语言
作者
Vinay Edula,Kalyan Ammisetty,Aakash Kotha,Deepthi Ravipati,Aakashnag Davuluri,P. Arulmozhivarman
标识
DOI:10.1109/icccnt56998.2023.10307854
摘要
QR codes have become increasingly popular in various applications, such as inventory management, advertising, and payment systems. However, detecting and decoding QR codes from blurry or unclear images can be challenging due to various factors, such as low image resolution, noise. The research aims to propose an efficient approach for detecting and decoding QR codes from unclear images using the You Only Look Once (YOLO) object detection model and deep super-resolution techniques implemented through Real-ESRGAN [1]. Authors trained and evaluated five different YOLOv8 object detection models with varying sizes and parameters to detect and crop the QR code from the input image. Based on the evaluation results, authors selected the model which achieved the best performance in terms of precision and recall. The cropped QR code based on YOLOv8 bounding box, is then enhanced using real-ESRGAN [1], which applies deep super-resolution techniques to improve the quality of the image. Finally, the enhanced QR code is decoded using three QR code utility libraries in Python, namely pyzbar, zxing, and opencv, which are executed in parallel using multithreading for efficient retrieval in less time. The results after testing demonstrates the effectiveness of proposed approach in accurately detecting and decoding QR codes from blurry or unclear images. The proposed approach can be used in various fields, such as inventory management, advertising, and payment systems, to provide a seamless user experience.
科研通智能强力驱动
Strongly Powered by AbleSci AI