An object detection based automated system for trash and hanging wire detection to enhance cleanliness and safety

Abstract

The present invention is related an object detection based automated system for trash and hanging wire detection to enhance cleanliness and safety. The invention presents a novel object-based detection system designed for trash detection and hanging wire identification to improve classroom aesthetics and cleanliness. The system uses object detection and contour analysis techniques to identify waste and hanging wires in classroom and real-world scenarios. It is implemented as a computer vision-based model, using a combination of shape-matching algorithms and boundary detection techniques running on jetson nano for real-time performance and high-performance speed. A key novelty of this system is real-time classroom maintenance monitoring using a hybrid system combining object detection and contour-based wire analysis to minimize false positives and notifies appropriate personnel in real time for efficient resolution. For trash detection, the system uses feature extraction and classification approach combine with multi-scale object recognition and intersection-over-union (IoU)-based optimization. Another novel contribution is a safety monitoring module that differentiates hanging electrical wires from other linear objects using chamfer distance, modified Hausdorff distance, and curvature similarity metrics. The system offers a real-time, low-power solution, making it ideal for applications in waste management, electrical safety monitoring, smart classrooms, or infrastructure development.