TSViT slightly outperforms 3D U-Net on crop segmentation from Sentinel-2 time series while VistaFormer offers the best efficiency-performance balance.
Al Rahhal, Reham Al Dayil, and Naif Al Ajlan
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Comparative Study of Transformer and Convolutional Models for Crop Segmentation from Satellite Image Time Series
TSViT slightly outperforms 3D U-Net on crop segmentation from Sentinel-2 time series while VistaFormer offers the best efficiency-performance balance.