Abstract
Observing spartina alterniflora (S.alterniflora) with multi-temporal remote sensing data aids in comprehending its development and spread in wetland ecosystems, thereby facilitating the formulation of effective strategies for its containment and control. Unsupervised Domain Adaptation (UDA) techniques uncover its spatio-temporal patterns, but most methods miss critical domain and class differences, whereas Intermediate Domain Prototype Class-level Learning Network (IDCNet) addresses these gaps. IDCNet generates class prototypes based on intermediate domain features, incorporating inter-class information for more accurate distribution alignment. Experimental results on two cross-year multi-spectral datasets demonstrate that the proposed IDCNet outperforms several state-of-the-art UDA methods.

This publication has 15 references indexed in Scilit: