DIGITAL BUSINESS CARD (DiNa) APPLICATION USING CNN ALGORITHM AND OCR TECHNOLOGY AS A FORMAL INTRODUCTION SUGGESTION

Abstract

The rapid development of technology makes digital business cards increasingly the first choice as a formal introduction tool that is more environmentally friendly, reducing dependence on the use of paper and ink. In addition to serving as a formal means of introduction, digital business cards are also an effective medium for conveying crucial information about an individual or company. The implementation phase of this application involves the utilization of Optical Character Recognition (OCR) as the main feature, with image pre-processing as a key step to improve reading accuracy, including noise reduction, data normalization, and compression. The process of optical scanning and location segmentation is the main foundation in processing data from Business Card images. The next step includes feature representation and extraction using TensorFlow's OCR technique to process the data efficiently. The integration of the OCR model into the API allows Kotlin-based mobile applications to communicate directly with the OCR model, providing real-time character recognition. The first trial aims to evaluate the accuracy and time taken by the OCR feature in recognizing each text on the Business Card. Tensorflow and Easy-OCR models with 41.86% accuracy were used for object detection and optical character recognition, resulting in a system that is efficient, responsive, and allows model updates without interrupting the main functionality of the application.  The app successfully combines eco-friendly aspects with advanced technology, creating a modern solution to meet the needs of effective formal introductions. Thus, this Digital Business Card application is not only an eco-friendly alternative, but also realizes efficiency in retrieving identification information directly through a mobile platform.