TransAdapter is a Swin Transformer architecture with graph-based domain discrimination, entropy-reweighted dual attention, and cross-feature transforms that reports state-of-the-art unsupervised domain adaptation results on Office-31, Office-Home, VisDA-2017, and DomainNet.
Energy-based self-training and normalization for unsupervised domain adaptation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
TransAdapter: Vision Transformer for Feature-Centric Unsupervised Domain Adaptation
TransAdapter is a Swin Transformer architecture with graph-based domain discrimination, entropy-reweighted dual attention, and cross-feature transforms that reports state-of-the-art unsupervised domain adaptation results on Office-31, Office-Home, VisDA-2017, and DomainNet.