Study of Handwriting Recognition Implementation in Data Entry of Survei Angkatan Kerja Nasional (SAKERNAS) using CNN
The use of Paper and Pencil Interviewing (PAPI) at BPS requires manual data entry that cannot be separated from the human ability to recognize handwriting. For computers, handwriting recognition is complex work that requires complex algorithms. Convolutional Neural Network (CNN) is an algorithm that can accommodate the complexity of handwriting recognition. This research intends to conduct a study on the implementation of the handwriting recognition model using CNN in recognizing handwriting on the PAPI questionnaire in data entry activities. Handwriting recognition model was built using the EMNIST dataset separately according to its character type and provides 89% accuracy for characters in the form of letters and numbers, 95% for characters in the form of letters, and 99% for characters in the form of numbers. Implementation of the handwriting recognition on the questionnaire image shows good results with 83.33% accuracy. However, there are problems found in the process of character segmentation where characters are not segmented correctly because the line of writing continues on the character that should be separated and disconnected characters when they should be joined. The result obtained in this study is expected to be a consideration regarding the entry method data used by BPS later.