CLASSIFICATION OF GUIDING BLOCKS FOR THE VISUALLY IMPAIRED USING HOG AND SVM

Nurul Fathanah Mustamin; Muhammad Afief Ma’Ruf; Rafif Taufiqurrahman

doi:10.20527/jtiulm.v8i1.151

Nurul Fathanah Mustamin Universitas Lambung Mangkurat
Muhammad Afief Ma’Ruf Universitas Lambung Mangkurat
Rafif Taufiqurrahman Universitas Lambung Mangkurat

DOI: https://doi.org/10.20527/jtiulm.v8i1.151

Keywords: Guiding Block, Computing Methodologies, HOG, Machine learning, Disability, Image Processing, Image Classification, Blind, SVM

Abstract

All citizens, including those with physical limitations, should be able to use public facilities and services without issue. The principles of accessibility—convenience, safety, usability, and independence—must be followed if we are to achieve social welfare in every area of life. For those with mobility or sight impairments, a guiding block (also known as a guiding tile or warning tile) can serve as an important indication. Regrettably, not all visually impaired people make use of this function in its current setup. Use a Support Vector Machine (SVM) and a His-togram of Oriented Gradient (HOG) to extract features for use in Guidance block image categorization. At a threshold of 75, the guiding blocks' classification accuracy is 99.51 percent within 30 cm, 74.811 percent within 50 cm, and 82.221 percent within 80 cm.

Author Biographies

Nurul Fathanah Mustamin, Universitas Lambung Mangkurat

Muhammad Afief Ma’Ruf, Universitas Lambung Mangkurat

Rafif Taufiqurrahman, Universitas Lambung Mangkurat