JSMNET: IMPROVING INDOOR POINT CLOUD SEMANTIC AND INSTANCE SEGMENTATION THROUGH SELF-ATTENTION AND MULTISCALE FUSION

This study produced the first sub-meter map of Spartina alterniflora in mainland China for 2020. Compared to the 10 m product, the sub-meter map achieved a 14.60% increase in overall accuracy, revealed a 57.73% discrepancy in spatial distribution, and detected 17 times more individual patches. The improved sensitivity to small patches significantly enhances early detection of Spartina alterniflora invasion. This finer mapping further refines soil carbon estimates and informs wetland management.