Design and Development of Model to Detect Garbage from CCTV

Abstract

Urban waste management is a pressing concern globally, necessitating innovative solutions for effective sanitation and sustainability. In response, this project introduces a Garbage Detection System leveraging deep learning techniques and web-based monitoring for real-time surveillance of city-wide Garbage accumulation. Utilizing the YOLOv7 object detection model, the system achieves an accuracy rate of 81% in identifying Garbage objects across diverse environmental conditions and camera angles. The accompanying web application dashboard provides intuitive access to live camera feeds, empowering stakeholders with proactive monitoring capabilities and facilitating prompt response to sanitation issues. Integration with MongoDB ensures efficient data storage and retrieval, enabling seamless access to historical Garbage detection data for analysis and optimization. The iterative refinement process, guided by periodic model updates and user feedback loops, contributes to continuous improvement in system performance. Through its comprehensive approach, this project demonstrates the feasibility and effectiveness of utilizing advanced computer vision technology for enhancing urban cleanliness and sustainability. Future endeavours will focus on further enhancing accuracy, scalability, and integration with existing waste management infrastructure to realize holistic urban waste management solutions. Key Words: YOLOv7, neural network, object detection, deep learning, web-based monitoring