ULFine combines confidence-aware text-prototype fitting with fused linear and similarity logits to make CLIP-based long-tailed semi-supervised learning more accurate and cheaper.
Dynamic learnable logit adjustment for long-tailed visual recognition,
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
-
ULFine: Unbiased Lightweight Fine-tuning for Foundation-Model-Assisted Long-Tailed Semi-Supervised Learning
ULFine combines confidence-aware text-prototype fitting with fused linear and similarity logits to make CLIP-based long-tailed semi-supervised learning more accurate and cheaper.