<![CDATA[Flooding is the most frequent natural disaster in Indonesia, with BNPB data showing a sharp increase since 2016, reaching 1,794 incidents in 2021. In 2024, 1,420 flood cases were reported, the majority of which were caused by waste accumulation in river channels. One example is the flood in Gadingrejo Village, Central Java, which submerged 100 houses due to waste obstructing the river flow. This issue motivated the development of RiverEye, an intelligent system based on CCTV and anomaly detection to prevent floods caused by waste in urban river channels. The system is designed using Raspberry Pi cameras, a buzzer as an early warning alarm, and a mini-computer running Artificial Intelligence (AI) models. The research methodology integrates YOLOv8 for object detection of waste and humans, MediaPipe Pose for detecting littering gestures, and face recognition to identify the perpetrators. The system includes a Flask-based monitoring dashboard that displays real-time detection results and a WhatsApp bot for automatic reporting. Testing was conducted on five main functions, achieving an average success rate of 89%, including pose detection 93%, object detection 90%, face recognition 83%, alarm 80%, and WhatsApp bot integration 100%. The findings demonstrate that RiverEye can detect littering behavior quickly and accurately, providing early warnings of potential river obstructions. The system has the potential to be applied as an effective, efficient, and environmentally friendly AI-based disaster mitigation tool. Further research is recommended to expand the testing area, increase the river visual dataset, and develop flood prediction features based on historical data for sustainable implementation.]]>
Copyrights © 2025
