Machine Learning Model for Baybayin (Alibata) To Tagalog Group of Words Level Transliteration
Vince Louies A. Alcantara, Hannah Michaela P. Dionida, Meo Vincent C. Caya · 2024
This study aims to develop an operational prototype that would perform Baybayin character detection and translate the Baybayin Phrases to Tagalog Phrases. This study describes an approach based on INCEPTION V4 Algorithm that utilizes KERAS for implementing neural networks in character recognition. By utilizing INCEPTION V4 Algorithm, the system yielded an accuracy of 98.69%, when recognizing Baybayin Characters, and 97.22% when recognizing Baybayin Phrases. By performing testing on the system, it was observed that there were some misclassifications while recognizing some of the characters and phrases, and it was observed that the factors involved in the failure of the recognition is the lighting on the environment, the distortion on the way the image was taken, and the unreadable handwriting. In the end, it was inferred that the recognition and transliteration of the Baybayin characters and phrases using the INCEPTION V4 algorithm produced a very high accuracy percentage that surpasses the expectation of the researchers, when recognizing and transliterating Baybayin Characters and Phrases.