Ekstraksi Data E-KTP untuk Pencatatan Peserta Vaksinasi Covid-19 dengan Metode YOLO V7 dan Paddle OCR Berbasis Android

Abstract

The advancement of information technology has significantly contributed to assisting human tasks. For instance, the manual data recording process in health centers indicates that the use of technology in Indonesia is not evenly distributed, especially during the COVID-19 vaccination process some time ago. As a result, the data recording process takes longer, causing delays in the vaccination administration process that should be quick. Therefore, this research has an aim for to detected and extracted data from the electronic Kartu Tanda Penduduk (e-KTP) for the COVID-19 vaccination participant recording process at health centers by implementing the You Only Look Once version 7 (YOLOv7) method and the Paddle OCR library, implemented on the Android system. In this study, 800 e-KTP image data were used, and two types of tests were conducted. First, the detection of all attributes on the e-KTP achieved excellent accuracy, specifically 98.19%. Second, the truth extraction test of e-KTP attribute data by the PaddleOCR library resulted in equally excellent accuracy, reaching 94.45%. Thus, the use of PaddleOCR with the YOLOv7 detection approach implemented on the Android system can be a solutive effort to modernize administrative governance in various public service sectors, especially in healthcare institutions.