CropNet, a lightweight CNN jointly convolving spectral and temporal dimensions, learns invariant crop signatures from multispectral time series and outperforms larger models under geographic domain shifts on the new CropGlobe benchmark spanning eight countries.
Visualizing data using t-sne
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Invariant Features for Global Crop Type Classification
CropNet, a lightweight CNN jointly convolving spectral and temporal dimensions, learns invariant crop signatures from multispectral time series and outperforms larger models under geographic domain shifts on the new CropGlobe benchmark spanning eight countries.