Classification and Detection of Visually Impaired Eyes Utilizing MTCNN and FaceNet Algorithm

Visually impaired children need information in different forms that are more accessible. They use alternative methods such as braille or audio to get information. However, they still find it challenging to get information in the school environment, such as class schedules, announcements, and changes in school activities. This study leverages neural network technology for the detection and classification of faces as visually impaired or non-visually impaired, employing MTCNN and FaceNet technologies. The system designed has demonstrated the ability to accurately identify faces, achieving a notable 91% accuracy rate during model training. This training involved testing on 30 distinct subjects and yielded considerable success. A total of 3360 data points were utilized for the training phase, while 840 were employed for model evaluation. The application of this technology harbors the promise of enhancing the inclusivity and educational experiences of visually impaired children in academic settings.