Swacch Suraksha - AI Enabled Smart Surviellance System for Anti - Littering Enforcement

The development of the automated object violation detection and alert system illustrates the powerful capabilities of integrating modern computer vision techniques, machine learning models, and real-time communication technologies into a unified intelligent surveillance framework. Through the combined use of YOLOv5 for high-accuracy object detection, LBPH for reliable and lightweight face recognition, IoU-based tracking for continuous monitoring of object movement, and rule-based analysis for detecting threshold violations, the system successfully identifies actions such as dropping or throwing bags, cups, and bottles in real time. The addition of instant SMS alerting through cloud-based communication further enhances the system’s practical utility by ensuring immediate accountability, reducing human oversight requirements, and reinforcing behavioral discipline among monitored individuals. The Flask-based web interface, paired with an SQLite database, provides an accessible and efficient platform for administrators to manage user registration, face training, and live monitoring. Experimental results confirm that the system performs consistently under varied environmental conditions and achieves stable detection, tracking, and recognition performance without requiring specialized GPU hardware. Overall, the project demonstrates that AI-driven surveillance systems can significantly improve monitoring efficiency, reduce manual intervention, and strengthen rule enforcement in public, institutional, and industrial environments. This work also highlights the potential for future expansion into multi-camera setups, deeper neural face recognition models, enhanced behavioral analytics, cloud storage integration, and large-scale deployments, making the system an adaptable and scalable solution for next-generation smart security infrastructures.

[1]             L. Ali, H. Baig, A. Khan and M. A. Khan, “Development of YOLOv5-Based Real-Time Smart Vision-Based Lab-Safety Monitoring System,” Sensors, vol. 22, no. 22, 2022. [2]             M. Córdova, J. A. Gutierrez, and A. D. P. Moreno, “Litter Detection with Deep Learning: A Comparative Study,” Sensors, vol. 22, no. 2, 2022. [3]             N. Zhao, X. Chen and Y. Zhang, “CMCA-YOLO: A Study on a Real-Time Object Detection Model with Context and Multi-Channel Attention,” Electronics, vol. 13, no. 8, 2024. [4]             G. Liu, X. Wang and H. Li, “Lightweight object detection algorithm for robots with improved YOLOv5,” Information Processing in Industry (journal), 2023. [5]             Ultralytics, “YOLOv5 — Ultralytics Documentation and Models,” Ultralytics docs / GitHub (model documentation and codebase), 2022–2024. [6]             “What is YOLOv5: A deep look into the internal features of YOLOv5,” arXiv preprint, 2024. [7]             M. Córdova et al., “Litter Detection with Deep Learning: A Comparative Study” (extended dataset & methods), Sensors, 2022 — (useful for dataset choices and evaluation metrics). [8]             J. Bao, “IOU-Guided Siamese Tracking,” Mathematical Problems in Engineering, 2021. [9]             K. Yang et al., “IASA: An IoU-aware tracker with adaptive sample assignment,” Journal / Elsevier (2023) — IoU-aware tracking improvements. [10]         “Real-Time Object Tracking With YOLOv5 and Recurrent Network on a Hardware Platform,” research article (2024) — integration of YOLOv5 + tracking for embedded platforms.