Per-pixel loss reweighting across stages, inspired by AdaBoost, improves gland segmentation accuracy over fixed-attention and non-boosted multi-stage baselines on a private dataset.
Going deeper with convolutions,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
AttentionBoost: Learning What to Attend by Boosting Fully Convolutional Networks
Per-pixel loss reweighting across stages, inspired by AdaBoost, improves gland segmentation accuracy over fixed-attention and non-boosted multi-stage baselines on a private dataset.