SITS-Extreme detects extreme events by learning patch representations from a satellite image time series via autoencoding with contrastive and consistency losses, then thresholding the mean cosine distance between pre- and post-disaster images.
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
-
Leveraging Satellite Image Time Series for Accurate Extreme Event Detection
SITS-Extreme detects extreme events by learning patch representations from a satellite image time series via autoencoding with contrastive and consistency losses, then thresholding the mean cosine distance between pre- and post-disaster images.