Communication Dans Un Congrès Année : 2025

Adapting Without Seeing: Text-Aided Domain Adaptation for Adapting CLIP-like Models to Novel Domains

Résumé

This paper addresses the challenge of adapting large vision models, such as CLIP, to domain shifts in image classification tasks. While these models, pre-trained on vast datasets like LAION 2B, offer powerful visual representations, they may struggle when applied to domains significantly different from their training data, such as industrial applications. We introduce TADA, a Text-Aided Domain Adaptation method that adapts the visual representations of these models to new domains without requiring target domain images. TADA leverages verbal descriptions of the domain shift to capture the differences between the pre-training and target domains. Our method integrates seamlessly with fine-tuning strategies, including prompt learning methods. We demonstrate TADA's effectiveness in improving the performance of large vision models on domain-shifted data, achieving state-of-the-art results on benchmarks like DomainNet.
Fichier principal
Vignette du fichier
icassp_2025.pdf (7.93 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04889885 , version 1 (15-01-2025)

Identifiants

  • HAL Id : hal-04889885 , version 1

Citer

Louis Hémadou, Héléna Vorobieva, Ewa Kijak, Frédéric Jurie. Adapting Without Seeing: Text-Aided Domain Adaptation for Adapting CLIP-like Models to Novel Domains. IEEE International Conference on Acoustics, Speech, and Signal Processing, Apr 2025, Hyderabad, India. ⟨hal-04889885⟩
0 Consultations
0 Téléchargements

Partager

More