The “AquaVision AI” project is dedicated to creating an accessible system for the early detection of environmental issues in water bodies using multispectral imaging and artificial intelligence. The author developed an optical system with interchangeable light filters, image analysis software, and trained the YOLOv26m model to detect signs of pollution, eutrophication, and debris accumulation based on changes in the spectral characteristics of the water.
The technology operates in several stages: first, the AI analyzes photos and videos from open sources and surveillance cameras; then, suspicious areas are photographed again using multispectral cameras or drones. The system can automatically detect anomalies, determine the coordinates of problem areas, generate PDF and JSON reports, and transmit data to environmental agencies.
The main advantage of the system is its combination of high accuracy, autonomous AI processing, and low cost – about $445 per monitoring module-making it accessible to communities, environmental activists, and local water body monitoring initiatives.
