Intelligent Extraction of Urban Vegetation Information Based on Segment Anything Model and Residual Neural Network Model
Source abstract
Abstract
To effectively support urban ecological environment assessment and gross ecosystem product (GEP) calculation, we used Beijing-2 and Beijing-3 satellite remote sensing images with a resolution of 0.8 m as data sources and a self-made urban vegetation sample dataset, and proposed a model combining the Segment Anything Model (SAM) and residual neural network (ResNet) for vegetation extraction.This method effectively combined the segmentation ability of SAM with the classification and extraction ability of the ResNet model.The model training results showed that the MIoU, MPA, and ACC of this method reached 77.27, 84.74, and 85.08%, respectively, which were slightly better than those of the ResNet model, and it can accurately segment and extract the microvegetation pattern in a complex urban environment.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Remote Sensing and Land Use · Soil and Land Suitability Analysis · AI and Multimedia in Education
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References
Reference metadata was not supplied for this record.