Year: 2024
Type: article
Abstract: In response to safety and security, accurate detection and alert systems for object identification in public spaces are crucial for enhancing surveillance for the visually impaired. Current object recognition algorithms face challenges due to lighting and spatial variations, resulting in inconsistent performance and lack of tailored features like color-based alerts for identification. Our study introduces an innovative algorithm that extracts invariant facial features through a unique segmentation method and K-Means clustering, significantly improving the reliability of face mask detection and enabling precise color recognition for visually impaired individuals through sRGB values and text-to-speech technology. Our research extends to evaluating our algorithm against leading object detection models like MobileNet SSD and Faster R-CNN, achieving a high weighted-average F1-score of 0.9131, showcasing the effectiveness of our approach. Future investigations will focus on enhancing color identification algorithms and expanding auditory signals to cover a wider range of objects, ultimately providing visually impaired individuals with a more comprehensive understanding of their surroundings for safer navigation.
Authors: Harshad N. Lokhande, Sanjay R. Ganorkar
Institution: Savitribai Phule Pune University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
SDG: Reduced inequalities
Open Access status: diamond
Funder: Savitribai Phule Pune University
OpenAlex ID: W4405194161