Cloud storage is an inconsistently rising technology that is used globally. Financial organizations and departments are the ones that most frequently need to decode the QR codes of the documents. They do it manually, which is time-consuming as well as cumbersome. QR code detection is a challenging problem statement with subdomains of manual intervention and error. The idea of the approach should be with a design of automation having accurate detection and decoding of a QR code. In this paper, the technology stack includes a YOLO (You look only once) object detection scheme where a darknet is integrated to detect a QR code in an image or a pdf. The system design is robust in nature and deals with various formats and a trained model that knows to detect the QR code from that document. In this project, image quality is also addressed and improved using GAN (Generative Adversarial Network). The best quality image is selected, and a trained model will detect the QR code and crop it so the frame will be a lossless object. The final step is decoding the QR code using frameworks like pyzbar, OpenCV, qrtools, and PIL. The combination of these terms and frameworks turned out to be an integrated, scalable, and ready-to-use product. The stepwise explanation is given, and, in each section, there are various frameworks and tools used. The research on detecting the QR code is reflected in this paper's multiple experiments on the documents.

Cloud storage, YOLO (You look only once), GAN (Generative Adversarial Network), pyzbar, OpenCV, and PIL, qrtools