Penerapan Optical Character Recognition Dalam Fitur Pemindaian Kartu Tanda Penduduk Pada Aplikasi Cash For Work

Abstract

Cash for Work (CFW) is part of the United Nations Development Programme (UNDP) initiative aimed at assisting communities affected by disasters or conflicts through the provision of employment opportunities. However, the lengthy registration process poses a significant challenge, especially with the increasing number of applicants. A large volume of applicants results in longer registration times, as facilitators must manually input data into the application one by one. To address this issue, Optical Character Recognition (OCR) technology was introduced, enabling facilitators to scan identification cards (KTP) to expedite the registration process and enhance accessibility for the community. The development of the OCR technology utilized Convolutional Neural Network (CNN), which has been proven effective in recognizing handwritten or printed characters from images. The CNN model was trained using a publicly available dataset containing alphanumeric characters and was tested on KTP images to identify characters individually through the sliding window method. This study produced an OCR model based on CNN with an accuracy rate of 92% in character extraction from KTP images when applied to the CFW program. This technology significantly improved the efficiency of the registration process, enabling a larger number of applicants to register more quickly and conveniently.