Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM62948
Volume:
10 |
5 |
May 2026
Pages:
1-7
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Computer Science and Engineering
Keshav Memorial Institute of Technology
Computer Science and Engineering
Keshav Memorial Institute of Technology
Computer Science and Engineering
Keshav Memorial Institute of Technology
Computer Science and Engineering
Keshav Memorial Institute of Technology
Computer Science and Engineering
Keshav Memorial Institute of Technology
Computer Science and Engineering
Keshav Memorial Institute of Technology
Computer Science and Engineering
Keshav Memorial Institute of Technology
Visual impairment restricts independent navigation, object recognition, text access, and timely hazard awareness in everyday indoor and outdoor environments. Conventional aids such as white canes, screen readers, and basic object detectors provide partial support, but they often lack contextual scene understanding, video-based interpretation, interactive follow-up assistance, and persistent activity history. This paper presents a mobile-based Visual Aid for Visually Impaired People that integrates Vision-Language Models (VLMs), Retrieval-Augmented Generation (RAG), hazard detection, and accessible audio feedback within a scalable Flutter-Flask-MongoDB architecture. The proposed system enables users to capture images or videos through a mobile interface, processes visual inputs through a Flask middleware layer, generates context-aware captions using a BLIP-based vision-language pipeline, detects hazardous objects or unsafe surroundings, and responds to user queries using stored captions and interaction history as retrieval context. Video inputs are divided into frames at fixed intervals of approximately one to two seconds, enabling scene-level descriptions from dynamic environments. Processed media, generated captions, user questions, responses, and hazard events are stored in MongoDB for future review by users or caregivers. Experimental evaluation over representative assistive-use scenarios shows that the system achieves high semantic caption adequacy, reliable hazard detection, and practical response latency for real-time assistance. The work demonstrates how multimodal AI and mobile computing can be combined into a low-cost, user-centered assistive platform that improves safety, independence, and environmental awareness for visually impaired users.
KEYWORDS
This paper presented a mobile-based Visual Aid for Visually Impaired People that integrates VLM captioning, video frame interpretation, RAG-based conversation, hazard detection, MongoDB history storage, and accessible audio feedback. The system addresses limitations of existing assistive technologies by combining contextual scene understanding with interactive and safety-aware assistance. The prototype demonstrates practical latency, strong caption adequacy, reliable hazard alerting, and meaningful follow-up query handling across representative scenarios. Overall, the project shows that modern vision-language AI and mobile software engineering can produce a scalable, low-cost, and user-centered assistive platform capable of improving independence, safety, and situational awareness for visually impaired individuals.
[1] S. Bhat, P. Bhat, and S. V. Kolekar, "From Vision to Voice: A Multi-Modal Assistive Framework for the Physically Impaired," IEEE Access, vol. 13, 2025, doi: 10.1109/ACCESS.2025.3590237. [2] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding," in Proc. NAACL-HLT, Minneapolis, MN, USA, 2019, pp. 4171-4186, doi: 10.18653/v1/N19-1423. [3] A. Radford et al., "Learning Transferable Visual Models From Natural Language Supervision," in Proc. 38th Int. Conf. Machine Learning (ICML), 2021. [4] J. Li, D. Li, C. Xiong, and S. Hoi, "BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation," in Proc. 39th Int. Conf. Machine Learning (ICML), 2022. [5] P. Lewis et al., "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks," in Advances in Neural Information Processing Systems, vol. 33, 2020, pp. 9459-9474. [6] W. Kim, B. Son, and I. Kim, "ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision," in Proc. 38th Int. Conf. Machine Learning (ICML), 2021. [7] S. Antol et al., "VQA: Visual Question Answering," in Proc. IEEE Int. Conf. Computer Vision (ICCV), Santiago, Chile, 2015, pp. 2425-2433, doi: 10.1109/ICCV.2015.279. [8] A. Vaswani et al., "Attention Is All You Need," in Advances in Neural Information Processing Systems, 2017, pp. 5998-6008. [9] J. Redmon and A. Farhadi, "YOLOv3: An Incremental Improvement," arXiv:1804.02767, 2018. [10] OpenCV, "OpenCV-Python Tutorials," OpenCV Documentation. [Online]. Available: https://docs.opencv.org/
Publication Date
5/10/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
0
Total views since publication
PDF Downloads
6
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...