CLIPScope: Enhancing Zero-Shot OOD Detection with Bayesian Scoring
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915011353378816 |
|---|---|
| author | Fu, Hao Patel, Naman Krishnamurthy, Prashanth Khorrami, Farshad |
| author_facet | Fu, Hao Patel, Naman Krishnamurthy, Prashanth Khorrami, Farshad |
| contents | Detection of out-of-distribution (OOD) samples is crucial for safe real-world deployment of machine learning models. Recent advances in vision language foundation models have made them capable of detecting OOD samples without requiring in-distribution (ID) images. However, these zero-shot methods often underperform as they do not adequately consider ID class likelihoods in their detection confidence scoring. Hence, we introduce CLIPScope, a zero-shot OOD detection approach that normalizes the confidence score of a sample by class likelihoods, akin to a Bayesian posterior update. Furthermore, CLIPScope incorporates a novel strategy to mine OOD classes from a large lexical database. It selects class labels that are farthest and nearest to ID classes in terms of CLIP embedding distance to maximize coverage of OOD samples. We conduct extensive ablation studies and empirical evaluations, demonstrating state of the art performance of CLIPScope across various OOD detection benchmarks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_14737 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | CLIPScope: Enhancing Zero-Shot OOD Detection with Bayesian Scoring Fu, Hao Patel, Naman Krishnamurthy, Prashanth Khorrami, Farshad Computer Vision and Pattern Recognition Detection of out-of-distribution (OOD) samples is crucial for safe real-world deployment of machine learning models. Recent advances in vision language foundation models have made them capable of detecting OOD samples without requiring in-distribution (ID) images. However, these zero-shot methods often underperform as they do not adequately consider ID class likelihoods in their detection confidence scoring. Hence, we introduce CLIPScope, a zero-shot OOD detection approach that normalizes the confidence score of a sample by class likelihoods, akin to a Bayesian posterior update. Furthermore, CLIPScope incorporates a novel strategy to mine OOD classes from a large lexical database. It selects class labels that are farthest and nearest to ID classes in terms of CLIP embedding distance to maximize coverage of OOD samples. We conduct extensive ablation studies and empirical evaluations, demonstrating state of the art performance of CLIPScope across various OOD detection benchmarks. |
| title | CLIPScope: Enhancing Zero-Shot OOD Detection with Bayesian Scoring |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2405.14737 |