Two-Cell Contractions of Filipino Braille Recognition Using Extreme Learning Machine

Braille is a method of reading and writing letters and numbers in the shape of slightly raised dots extensively used by the blind. Filipino braille equivalent is mainly applied by the Filipino braille readers where there are unique braille characters compared to the international ones. However, only a few studies about it exist, and there is no practical device created exclusively for Filipino braille yet. This paper presents research and development to characterize Filipino two-cell contractions using image processing and extreme learning machine. A Raspberry Pi, 3 Model B, is used as the microcontroller of the prototype alongside an adaptive 8MP Raspberry Pi Noir camera for its image processing and deep learning. Python programming language is used to implement the system’s recognition alongside pre-existing libraries and scripts such as Scikit-Image, OpenCV, OBR, image processing, and image augmentation. Two hundred sixty samples were tested to know the system’s effectiveness, giving 76.54% recognition accuracy. Hence, the system manifests an acceptable method for recognizing Two-Cell Contractions of the Filipino Braille. In the future, the study is to be expected to have more datasets and applications of other Filipino Braille based on the DepEd Instruction Manual.