Konversi Lembar Jawaban Ujian Mahasiswa Menjadi Dokumen Digital Menggunakan Convolutional Neural Network (CNN) dan Long Short-Term Memory (LSTM)

Abstract

This application presents the process of recognizing words that can be done by automatically recognizing and converting words in the form of answer sheets from text documents (hardcopy) into document files (softcopy). Furthermore, preprocessing is done to simplify the segmentation stage. Image segmentation is one of the stages of image processing to divide several objects according to predetermined criteria. With the Optical Character Recognition method, image segmentation can be performed. Optical Character Recognition (OCR) aims to identify a point in the image based on the brightness level. The application of Convolutional Neural Network and Long Short-Term Memory (LSTM) method is applied to process small information from images that match the image templates. The application will detect whether the entered image can be tested clearly or not. The results of the training will be carried out by validating each answer sheet input. Based on the tests that have been carried out, the results obtained from this application are an average of 80% of 90 data.