Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{197650,
        author = {Dr. MK Jayanthi Kannan and Paarth Juneja and Manav Tiwari and Kartik Modi and Hardik Jain and Sankalp Agnihotri and Sarthak Tiwari},
        title = {RailVLM: An Integrated YOLOv8 and Vision Language Model Framework for Explainable Real-Time Railway Track Fault Detection and Maintenance Decision Support},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {7808-7816},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=197650},
        abstract = {RailVLM is an intelligent real-time railway inspection system designed to detect faults in railway tracks using advanced Artificial Intelligence techniques. The system combines high-speed object detection using YOLOv8 with Vision Language Models (VLMs) such as Moondream2 or API-based models like Gemini to provide explainable and human-readable diagnostics. It automates the detection of defects like cracks, missing fasteners, and obstacles, reducing dependency on manual inspection methods. The platform also provides real-time alerts and generates detailed maintenance reports, ensuring faster and more accurate decision-making. By integrating computer vision with explainable AI, RailVLM enhances railway safety, minimizes human error, and improves overall inspection efficiency. RailVLM introduces an automated, real-time railway inspection system that addresses these limitations through a synergistic combination of computer vision and explainable artificial intelligence. The system architecture comprises three core components: YOLOv8-based Fault Detection: High-speed object detection capable of identifying four primary defect categories in real-time video streams at 40+ frames per second. Vision Language Model Integration: Moondream2 (local, privacy-preserving) and Gemini API (cloud, high-capacity) generate natural language explanations for detected anomalies, providing actionable maintenance insights. Centralized Decision Support Platform: Web-based dashboard delivering real-time alerts, geotagged fault visualization, automated report generation, and maintenance prioritization recommendations.},
        keywords = {RailVLM, YOLOv8, Vision Language Model, Railway Fault Detection, Explainable AI},
        month = {April},
        }