A multi-task vision transformer trained on NAIP imagery and 3DEP lidar estimates 0.5m canopy height with lower median error than three published CHM models across most western US ecoregions.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
VibrantVS: A high-resolution multi-task transformer for forest canopy height estimation
A multi-task vision transformer trained on NAIP imagery and 3DEP lidar estimates 0.5m canopy height with lower median error than three published CHM models across most western US ecoregions.