Optical Character Recognition of Baybayin Writing System using YOLOv3 Algorithm

In the Philippines, Baybayin is one of its writing systems that originated in pre-Hispanic colonialism. The centuries-old writing system gained attention and popularity, which later turned into an approved bill in 2018. The recent development of research aimed at translating Baybayin characters into Alphabets, the globally recognizable writing system, uses Artificial Intelligence or A.I. Different researchers have developed an optical character recognition system for the Baybayin script but are incapable of translating multiple characters in single image and are all using object classification algorithms. Therefore, there is a need for a system using a YOLOv3 based CNN architecture capable of recognizing Baybayin scripts in word form. Using the YOLOv3 algorithm, the system was able to achieve an accuracy of 98.92%. It was observed that some of the misclassifications are due to distorted or illegible handwriting. It can be concluded that the optical character recognition of Baybayin characters using the YOLOv3 algorithm is of high accuracy when it comes to detecting and classifying Baybayin characters.