Automatic Features Detection in a Fluvial Environment through Machine Learning Techniques Based on UAVs Multispectral Data

The present work aims to demonstrate how machine learning (ML) techniques can be used for automatic feature detection and extraction in fluvial environments. use of photogrammetry algorithms has improved the understanding both environmental anthropic issues. developed methodology was applied considering acquisition multiple photogrammetric images thanks unmanned aerial vehicles (UAV) carrying multispectral cameras. These surveys were carried out Salbertrand area, along Dora Riparia River, situated Piedmont (Italy). authors an algorithm able identify detect water table contour concerning landed areas: classification ML found a valid identification different patterns (water, gravel bars, vegetation, ground classes) specific hydraulic geomatics conditions. Indeed, RE+NIR data gave us sharp rise terms accuracy by about 11% 13.5% F1-score average values testing point clouds compared RGB data. obtained results led define new procedure with precise validity