Smart Vision: Real-Time Object Detection and Audio Assistance for the Visually Impaired Using TensorFlow and SSD MobileNet
Year: 2025
Type: article
Abstract: Visual impairment affects approximately 8.5 billion people worldwide, limiting their ability to interpret visual information daily. To address this challenge, we propose an Android-based object detection application, "Digital Eyes," which assists visually impaired individuals by detecting objects and converting them into audio feedback. The application utilizes TensorFlow’s SSD MobileNet model for real-time object recognition. It identifies 10 categories of objects, including chairs, bottles, laptops, and persons, and provides spoken feedback via Text-to-Speech (TTS) technology. The system performs all tasks locally, eliminating the need for an internet connection and making it accessible on most smartphones. SSD MobileNet was chosen due to its balance between speed and accuracy, achieving a mean Average Precision (mAP) of ~74 and a frame rate of 59 FPS, making it suitable for real-time applications. The proposed system enhances the independence of visually impaired users by identifying objects and reading text from the environment. The results demonstrate high accuracy in object detection and reliable audio feedback, enabling a smoother user experience.
Author: Diksha R. Pawar
Language: English
Subfield: Occupational Therapy
Field: Health Professions
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W4409090570