HAM trains per-domain CLIP encoders, enriches them with confident cross-domain samples, aligns their update directions, and merges them with redundancy trimming, reaching 79.0% average accuracy on five DG benchmarks with ViT-B/16.
Leveraging vision-language models for improving domain generalization in image classification
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Harmonizing and Merging Source Models for CLIP-based Domain Generalization
HAM trains per-domain CLIP encoders, enriches them with confident cross-domain samples, aligns their update directions, and merges them with redundancy trimming, reaching 79.0% average accuracy on five DG benchmarks with ViT-B/16.