Deteksi Simbol pada Ekspresi Matematika dengan Menggunakan Faster Region Convolutional Neural Network (Faster R-CNN)

Abstract

Mathematics is an essential subject in the field of education, yet many students, particularly college students, face difficulties in studying it. One of the main causes is a lack of understanding of mathematical symbols, which are often challenging to recognize through Optical Character Recognition (OCR) processes. This creates a challenge for students to search for and identify the symbols being utilized. Therefore, a detection system is necessary to tackle this problem. This study aims to detect mathematical symbols in mathematical expressions using the Faster Region Convolutional Neural Network (Faster R-CNN) method. The dataset utilized consists of mathematical expression data collected from the 15th International Conference on Document Analysis and Recognition (ICDAR) in 2019, along with supporting data from various sources related to the field of mathematics, totaling 19,600 images with 84 symbol classes. The research results demonstrate that the developed system successfully detects mathematical symbols with an accuracy rate of up to 91%. These findings indicate the system's ability to accurately recognize and identify mathematical symbols. In conclusion, the Faster R-CNN method has been successfully implemented to detect mathematical symbols in mathematical expressions.