Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2308.10450 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910360119803904 |
|---|---|
| author | Liu, Xinghong Zhou, Yi Zhou, Tao Feng, Chun-Mei Shao, Ling |
| author_facet | Liu, Xinghong Zhou, Yi Zhou, Tao Feng, Chun-Mei Shao, Ling |
| contents | Universal domain adaptation (UniDA) aims to address domain and category shifts across data sources. Recently, due to more stringent data restrictions, researchers have introduced source-free UniDA (SF-UniDA). SF-UniDA methods eliminate the need for direct access to source samples when performing adaptation to the target domain. However, existing SF-UniDA methods still require an extensive quantity of labeled source samples to train a source model, resulting in significant labeling costs. To tackle this issue, we present a novel plug-and-play classifier-oriented calibration (COCA) method. COCA, which exploits textual prototypes, is designed for the source models based on few-shot learning with vision-language models (VLMs). It endows the VLM-powered few-shot learners, which are built for closed-set classification, with the unknown-aware ability to distinguish common and unknown classes in the SF-UniDA scenario. Crucially, COCA is a new paradigm to tackle SF-UniDA challenges based on VLMs, which focuses on classifier instead of image encoder optimization. Experiments show that COCA outperforms state-of-the-art UniDA and SF-UniDA models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2308_10450 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | COCA: Classifier-Oriented Calibration via Textual Prototype for Source-Free Universal Domain Adaptation Liu, Xinghong Zhou, Yi Zhou, Tao Feng, Chun-Mei Shao, Ling Computer Vision and Pattern Recognition Universal domain adaptation (UniDA) aims to address domain and category shifts across data sources. Recently, due to more stringent data restrictions, researchers have introduced source-free UniDA (SF-UniDA). SF-UniDA methods eliminate the need for direct access to source samples when performing adaptation to the target domain. However, existing SF-UniDA methods still require an extensive quantity of labeled source samples to train a source model, resulting in significant labeling costs. To tackle this issue, we present a novel plug-and-play classifier-oriented calibration (COCA) method. COCA, which exploits textual prototypes, is designed for the source models based on few-shot learning with vision-language models (VLMs). It endows the VLM-powered few-shot learners, which are built for closed-set classification, with the unknown-aware ability to distinguish common and unknown classes in the SF-UniDA scenario. Crucially, COCA is a new paradigm to tackle SF-UniDA challenges based on VLMs, which focuses on classifier instead of image encoder optimization. Experiments show that COCA outperforms state-of-the-art UniDA and SF-UniDA models. |
| title | COCA: Classifier-Oriented Calibration via Textual Prototype for Source-Free Universal Domain Adaptation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2308.10450 |