Alt-Eye: A Smart Vision Assistance Device for the Visually Impaired Using Deep Learning

View at publisher Summarise with VIRAT
Authors: K. Kartheeban, Thummala Nithin Reddy, P. Naga Ramya Sri, Enjirapu Yuva Sekhar Nikhil Eswara Achari
Cited by: 0
References: 15

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper introduces Alt-Eye, a wearable smart vision assistance system that enables visually impaired individuals to recognize objects in real time, perform optical character recognition, detect currency, and understand contextual scenes. The system comprises a Raspberry Pi Zero 2W interfaced with a Pi Camera V1 (OV5647) mounted on custom 3D-printed spectacle frames, connected to Bluetooth earphones for audio feedback. Alt-Eye adopts an edge-cloud hybrid architecture: while images are captured at the edge device, inference is executed on the cloud infrastructure provided by Microsoft Azure running a deep learning multimodal fusion pipeline that merges the output of a visionlanguage model with large language model reasoning. The system has an end-to-end latency of 2.5-3 seconds and is able to identify objects, extract text from product labels including MRP and expiry dates, recognize Indian currency denominations, and generate human-understandable descriptions using speech synthesis. Power is supplied via a custom battery management system using a 3.7 V 1000 mAh lithium polymer battery integrated with a charging and 5 V boost circuit that enables portable operation for 8-10 hours. Experimental validation is performed by achieving reliable detection accuracy across common indoor scenarios and product recognition tasks and represents a significant advancement in assistive technology by delivering context-aware, multimodal intelligence in a lightweight, ergonomic wearable form factor.

Fields of Study & Subjects

Assistive Technology in Communication and Mobility Advanced Neural Network Applications Multimodal Machine Learning Applications

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1214–1222
Publication date
21 January 2026
Language
EN