Attention-guided cropping in raw and feature space plus soft matching improves CLIP zero-shot classification and out-of-distribution generalization by 0.3 to 6.2 points, with no training.
Food-101 -- mining discriminative components with random forests
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
From Local Details to Global Context: Advancing Vision-Language Models with Attention-Based Selection
Attention-guided cropping in raw and feature space plus soft matching improves CLIP zero-shot classification and out-of-distribution generalization by 0.3 to 6.2 points, with no training.