Unsupervised Domain Adaption Harnessing Vision-Language Pre-Training
IEEE Transactions on Circuits and Systems for Video Technology · 2024-09 · 34 (9) · 8201-8214
DOI: 10.1109/tcsvt.2024.3391304 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Circuits and Systems for Video Technology, 2024-09. Verfasst von Wenlve Zhou und 1 weiteren Autor:innen. Die Arbeit wurde bisher 25-mal zitiert (Crossref) und nennt 74 Literaturangaben. Gefördert von: National Key Research and Development Program of China, Basic and Applied Basic Research Foundation of Guangdong Province, Guangdong Provincial Key Laboratory of Human Digital Twin.
Abstract (Originalsprache)
This paper addresses two vital challenges in Unsupervised Domain Adaptation (UDA) with a focus on harnessing the power of Vision-Language Pre-training (VLP) models. Firstly, UDA has primarily relied on ImageNet pre-trained models. However, the potential of VLP models in UDA remains largely unexplored. The rich representation of VLP models holds significant promise for enhancing UDA tasks. To address this, we propose a novel method called Cross-Modal Knowledge Distillation (CMKD), leveraging VLP models as teacher models to guide the learning process in the target domain, resulting in state-of-the-art performance. Secondly, current UDA paradigms involve training separate models for each task, leading to significant storage overhead and impractical model deployment as the number of transfer tasks grows. To overcome this challenge, we introduce Residual Sparse Training (RST) exploiting the benefits conferred by VLP’s extensive pre-training, a technique that requires minimal adjustment (approximately 0.1%~0.5%) of VLP model parameters to achieve performance comparable to fine-tuning. Combining CMKD and RST, we present a comprehensive solution that effectively leverages VLP models for UDA tasks while reducing storage overhead for model deployment. Furthermore, CMKD can serve as a baseline in conjunction with other methods like FixMatch, enhancing the performance of UDA. Our proposed method outperforms existing techniques on standard benchmarks. Our code will be available at: https://github.com/Wenlve-Zhou/VLP-UDA.
Förderung
- National Key Research and Development Program of China
- Basic and Applied Basic Research Foundation of Guangdong Province
- Guangdong Provincial Key Laboratory of Human Digital Twin
- Guangzhou City Science and Technology Research Projects
- Jiangmen Science and Technology Research Projects
- Shaoguan Science and Technology Research Project
- Foshan Science and Technology Research Project
Zitieren
APA
Zhou, W., & Zhou, Z. (2024). Unsupervised Domain Adaption Harnessing Vision-Language Pre-Training. IEEE Transactions on Circuits and Systems for Video Technology, 34(9), 8201-8214. https://doi.org/10.1109/tcsvt.2024.3391304
Vancouver
Zhou W, Zhou Z. Unsupervised Domain Adaption Harnessing Vision-Language Pre-Training. IEEE Transactions on Circuits and Systems for Video Technology. 2024;34(9):8201-8214. doi:10.1109/tcsvt.2024.3391304
BibTeX
@article{Zhou2024,
author = {Zhou, Wenlve and Zhou, Zhiheng},
title = {Unsupervised Domain Adaption Harnessing Vision-Language Pre-Training},
journal = {IEEE Transactions on Circuits and Systems for Video Technology},
year = {2024},
volume = {34},
number = {9},
pages = {8201--8214},
doi = {10.1109/tcsvt.2024.3391304}
}
EUVIMED