Research map / Label-efficient and robust learning / Test-time adaptation
TTL: Test-time Textual Learning for OOD Detection with Pretrained Vision-Language Models
Jinlun Ye, Jiang Liao, Runhe Lai, Xinhua Lu, Jiaxin Zhuang, Zhiyong Gan, Ruixuan Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add L_OMB pseudo-label prompt learning (ablation) | MCM base detector · ImageNet-1k | FPR95: 42.77 -> 30.56 | Improved |
| add L_OMB pseudo-label prompt learning (ablation) | MCM base detector · CIFAR-100 | FPR95: 73.09 -> 14.40 | Improved |
| add OOD knowledge purification L_OKP (ablation) | TTL (L_OMB + OKB) · ImageNet-1k | FPR95: 18.40 -> 12.46 | Improved |
| add OOD knowledge purification L_OKP (ablation) | TTL (L_OMB + OKB) · CIFAR-100 | FPR95: 5.23 -> 2.36 | Improved |
| add OOD Textual Knowledge Bank (ablation) | TTL (L_OMB + L_OKP) · ImageNet-1k | FPR95: 24.59 -> 12.46 | Improved |
| add OOD Textual Knowledge Bank (ablation) | TTL (L_OMB + L_OKP) · CIFAR-100 | FPR95: 4.14 -> 2.36 | Improved |
| use proposed lowest-similarity OKB update instead of FIFO (ablation) | TTL · ImageNet-1k | FPR95: 14.69 -> 12.46 | Improved |
| use proposed OKB update instead of random replacement (ablation) | TTL · CIFAR-100 | FPR95: 29.06 -> 2.36 | Improved |
| initialize OOD prompt prefix and classname manually instead of randomly (ablation) | TTL · CIFAR-100 | FPR95: 45.23 -> 2.36 | Improved |
| initialize prefix with 'a photo of a' while keeping random classname (ablation) | TTL · CIFAR-100 | FPR95: 45.23 -> 45.29 | No clear effect |
| add a visual knowledge bank (TTL-V) | TTL · ImageNet-1k | FPR95: 12.46 -> 9.19 | Improved |
| stop OOD prompt updates after 1,280 test samples (ablation) | TTL · ImageNet-1k | FPR95: 12.46 -> 14.21 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- AdaNeg — AdaNeg: Adaptive Negative Proxy Guided OOD Detection with Vision-Language Models (NeurIPS 2024) · better on 4 of 4 shared results
- OODD — OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary (CVPR 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- The Illusion of Progress? A Critical Look at Test-Time Adaptation for Vision-Language Models (NeurIPS 2025 · similarity 0.89)
- Conjugated Semantic Pool Improves OOD Detection with Pre-trained Vision-Language Models (NeurIPS 2024 · similarity 0.89)
- Test-Time Adaptation of Vision-Language Models for Open-Vocabulary Semantic Segmentation (NeurIPS 2025 · similarity 0.88)
- Noisy Test-Time Adaptation in Vision-Language Models (ICLR 2025 · similarity 0.88)
- $\Delta \mathrm{Energy}$: Optimizing Energy Change During Vision-Language Alignment Improves both OOD Detection and OOD Generalization (NeurIPS 2025 · similarity 0.88)
- What Drives Test-Time Adaptation for CLIP? A Controlled Empirical Study from an Update Perspective (NeurIPS 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
