SWIFT adapts pre-trained pansharpening models to new sensors in about one minute for most models by sampling 3% of target data and updating roughly 30% of weights, matching full retraining performance.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SWIFT: A General Sensitive Weight Identification Framework for Fast Sensor-Transfer Pansharpening
SWIFT adapts pre-trained pansharpening models to new sensors in about one minute for most models by sampling 3% of target data and updating roughly 30% of weights, matching full retraining performance.