An extension of FamNet with extra loss achieves 1.96 MAE when counting machine parts, outperforming traditional image processing, instance segmentation, and standard density estimation baselines.
Vi- sion transformers for dense prediction
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
-
Counting Machine Parts
An extension of FamNet with extra loss achieves 1.96 MAE when counting machine parts, outperforming traditional image processing, instance segmentation, and standard density estimation baselines.