Knowledge distillation from a rigid-invariant 3D point cloud network into a regulated multi-view Transformer yields lower-error, faster wheat spike volume estimates from 2D images.
A tale of two features: Stable diffusion complements DINO for zero-shot semantic correspondence
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
3D Reconstruction and Knowledge Distillation to Improve Multi-View Image Models to Explore Spike Volume Estimation in Wheat
Knowledge distillation from a rigid-invariant 3D point cloud network into a regulated multi-view Transformer yields lower-error, faster wheat spike volume estimates from 2D images.