JSSFF improves remote sensing image captioning by fusing structural edge details with semantic features in an encoder-decoder model and using fairness-based beam search, outperforming baselines on quantitative and qualitative measures.
Correcting low-illumination images using multi-scale fusion in a pyramidal framework
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
JSSFF: A Joint Structural-Semantic Fusion Framework for Remote Sensing Image Captioning
JSSFF improves remote sensing image captioning by fusing structural edge details with semantic features in an encoder-decoder model and using fairness-based beam search, outperforming baselines on quantitative and qualitative measures.