A Behavior-Oriented Mobile System For Real-Time Waste Object Detection and Classification Using Ssd Mobilenet V2　　

Improper waste disposal remains a major contributor to environmental degradation, public health risks, and urban flooding, particularly in regions with limited waste-management literacy. This study proposes an application that leverages artificial intelligence–based waste object detection to strengthen sustainable environmental education and support early flood-risk prevention. Building upon the SSD MobileNet V2 architecture, the system performs real-time waste identification on mobile devices, enabling accessible and interactive learning for diverse communities across borders. The model reaches a Mean Average Precision (MAP) of 85.32% and runs on a mid-range smartphone in about 138 ms per frame. The detected waste classes are linked to educational modules that highlight proper disposal practices, environmental hygiene, and the health impacts of littering. To enhance community-level resilience, the system incorporates a lightweight information layer that allows users to report waste accumulation and potential drainage blockages as early indicators of flood hazards. By integrating AI-driven object detection, behavioral education, and community-based environmental monitoring, this work offers a scalable and inclusive approach to reducing littering behavior, improving public health awareness, and strengthening flood-risk mitigation through sustainable cross-border education.