Sight Assist: Advanced Deep Learning for Real-Time Object Identification and Assistance for the Blind

Abstract

Visual impairment significantly limits an individual's ability to identify surrounding objects and navigate safely in unfamiliar environments. Traditional assistive tools such as white canes and guide dogs provide basic mobility support but lack the capability to recognize and describe nearby objects. This paper presents Sight Assist, an advanced deep learning–based system designed for real-time object identification to assist visually impaired individuals [6], [8]. The proposed system utilizes a camera to capture live video input and processes each frame using an object detection model to recognize multiple real-world objects such as people, vehicles, animals, and everyday items. The system is developed using HTML, CSS, and JavaScript for the front-end interface, while Python with the Flask framework is used for backend processing and integration of the deep learning model. Detected objects are converted into voice feedback using a multilingual text-to-speech module that supports Tamil, English, and Hindi, enabling users to receive audio descriptions of objects in their preferred language. By combining real-time object detection [1], [2] with multilingual voice assistance, the system enhances environmental awareness and promotes independence for visually impaired users. Experimental results demonstrate that the proposed solution provides efficient and reliable real-time assistance for object identification in everyday environments.