CLIP-driven Outliers Synthesis for few-shot OOD detection
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917626930790400 |
|---|---|
| author | Sun, Hao He, Rundong Han, Zhongyi Lin, Zhicong Gong, Yongshun Yin, Yilong |
| author_facet | Sun, Hao He, Rundong Han, Zhongyi Lin, Zhicong Gong, Yongshun Yin, Yilong |
| contents | Few-shot OOD detection focuses on recognizing out-of-distribution (OOD) images that belong to classes unseen during training, with the use of only a small number of labeled in-distribution (ID) images. Up to now, a mainstream strategy is based on large-scale vision-language models, such as CLIP. However, these methods overlook a crucial issue: the lack of reliable OOD supervision information, which can lead to biased boundaries between in-distribution (ID) and OOD. To tackle this problem, we propose CLIP-driven Outliers Synthesis~(CLIP-OS). Firstly, CLIP-OS enhances patch-level features' perception by newly proposed patch uniform convolution, and adaptively obtains the proportion of ID-relevant information by employing CLIP-surgery-discrepancy, thus achieving separation between ID-relevant and ID-irrelevant. Next, CLIP-OS synthesizes reliable OOD data by mixing up ID-relevant features from different classes to provide OOD supervision information. Afterward, CLIP-OS leverages synthetic OOD samples by unknown-aware prompt learning to enhance the separability of ID and OOD. Extensive experiments across multiple benchmarks demonstrate that CLIP-OS achieves superior few-shot OOD detection capability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_00323 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | CLIP-driven Outliers Synthesis for few-shot OOD detection Sun, Hao He, Rundong Han, Zhongyi Lin, Zhicong Gong, Yongshun Yin, Yilong Computer Vision and Pattern Recognition Machine Learning Few-shot OOD detection focuses on recognizing out-of-distribution (OOD) images that belong to classes unseen during training, with the use of only a small number of labeled in-distribution (ID) images. Up to now, a mainstream strategy is based on large-scale vision-language models, such as CLIP. However, these methods overlook a crucial issue: the lack of reliable OOD supervision information, which can lead to biased boundaries between in-distribution (ID) and OOD. To tackle this problem, we propose CLIP-driven Outliers Synthesis~(CLIP-OS). Firstly, CLIP-OS enhances patch-level features' perception by newly proposed patch uniform convolution, and adaptively obtains the proportion of ID-relevant information by employing CLIP-surgery-discrepancy, thus achieving separation between ID-relevant and ID-irrelevant. Next, CLIP-OS synthesizes reliable OOD data by mixing up ID-relevant features from different classes to provide OOD supervision information. Afterward, CLIP-OS leverages synthetic OOD samples by unknown-aware prompt learning to enhance the separability of ID and OOD. Extensive experiments across multiple benchmarks demonstrate that CLIP-OS achieves superior few-shot OOD detection capability. |
| title | CLIP-driven Outliers Synthesis for few-shot OOD detection |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2404.00323 |