YCL Based Smart Glasses for Visually Impaired

Abstract

A key component of assistive technology that helps people with visual impairments access text in their daily lives is a text reader. An image-based text recognition tool for the blind is presented in this article. It takes pictures of the user's surroundings using a dual-camera module. For improving character detection and recognition in these developed smart glasses for the blind and the visually impaired, a hybrid algorithm called the YCL Character Recognition Algorithm that combines You Only Look Once (YOLO), Convolutional Recurrent Neural Networks (CRNN), and Long Short-Term Memory (LSTM) to balance out the drawbacks of each algorithm by learning from its advantages. The suggested YOLO-v8 model is used for real-time text object detection, CRNN is used to extract character features, and LSTM is used to enhance sequential character prediction. An audio output signal is given to the user by the image processing software after the visual data has been processed. These smart glasses have the benefit of being able to view characters from both close and far distances. A specially acquired dataset was used to evaluate the suggested YCL technique, which shows notable speed and accuracy gains over the traditional homogenous algorithms. The suggested method is successful in identifying words and characters and in delivering audio output for the blind and visually impaired, based on users’ surveys and experimental data.