CaneShield: Vision-based Full-Stack System for Real- Time Metal and Disease Detection in Sugarcane

Metal contamination in sugarcane processing industries poses significant risks to machinery, product quality, and operational safety. This paper presents a real-time metal detection and monitoring system developed using the MERN stack and integrated with an AIbased microservice. The proposed system provides continuous monitoring through an interactive dashboard that enables users to visualize system status and detection events in real time. It incorporates an alert mechanism for immediate notification of ferrous and non-ferrous metal detection, helping to prevent equipment damage and reduce operational downtime. To ensure secure access, the system employs JSON Web Token (JWT) authentication and password hashing techniques. Additionally, it maintains historical detection records and supports automated report generation for analysis and documentation. The modular architecture facilitates efficient communication between the frontend, backend, and AI service, enhancing scalability and flexibility. Experimental evaluation demonstrates the system's effectiveness in improving industrial safety, reliability, and operational efficiency. Furthermore, the proposed solution provides a foundation for future integration with hardware sensors and advanced machine learning models.