TransRefine: Transformer-augmented feature refinement for zero-shot scene classification in remote sensing images

Published: 01 Jan 2025, Last Modified: 01 May 2025Pattern Recognit. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Introduced semantic attributes for zero-shot remote sensing scene classification.•TransRefine is proposed to combine embedding and generative-based methods.•Semantic attribute-guided Transformer extracts attribute-based visual features.•Feature refinement obtains discriminative visual features for seen & unseen classes.
Loading