mlsanityData to 14 Oct 2026

Activation Matters: Test-time Activated Negative Labels for OOD Detection with Vision-Language Models

arXiv 2603.25250 ↗First posted 26 Mar 2026cs.CV cs.AI cs.LGcited by 2

On the map: Multimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP

Abstract (arXiv)

Out-of-distribution (OOD) detection aims to identify samples that deviate from in-distribution (ID). One popular pipeline addresses this by introducing negative labels distant from ID classes and detecting OOD based on their distance to these labels. However, such labels may present poor activation on OOD samples, failing to capture the OOD characteristics. To address this, we propose underlineTest-time underlineActivated underlineNegative underlineLabels (TANL) by dynamically evaluating activation levels across the corpus dataset and mining candidate labels with high activation responses during the testing process. Specifically, TANL identifies high-confidence test images online and accumulates their assignment probabilities over the corpus to construct a label activation metric. Such a metric leverages historical test samples to adaptively align with the test distribution, enabling the selection of distribution-adaptive activated negative labels. By further exploring the activation information within the current testing batch, we introduce a more fine-grained, batch-adaptive variant. To fully utilize label activation knowledge, we propose an activation-aware score function that emphasizes negative labels with stronger activations, boosting performance and enhancing its robustness to the label number. Our TANL is training-free, test-efficient, and grounded in theoretical justification. Experiments on diverse backbones and wide task settings validate its effectiveness. Notably, on the large-scale ImageNet benchmark, TANL significantly reduces the FPR95 from 17.5% to 9.8%. Codes are available at hrefhttps://github.com/YBZh/OpenOOD-VLMYBZh/OpenOOD-VLM.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Prompt learning and adaptation of CLIP, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 23 of its 86 references that are in the map. First posted on arXiv 26 Mar 2026.

Builds on

23 of 86 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025Noisy Test-Time Adaptation in Vision-Language Models20 Feb
  2. 2024Conjugated Semantic Pool Improves OOD Detection with Pre-trained Vision-Language Models11 Oct
  3. 2024Generalized Out-of-Distribution Detection and Beyond in Vision Language Model Era: A Survey31 Jul

Other references · 20

  1. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,877
  2. 2016A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networkscited by 2,385
  3. 2017Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networkscited by 1,328
  4. 2018Deep Anomaly Detection with Outlier Exposurecited by 942
  5. 2019AugMix: A Simple Data Processing Method to Improve Robustness and Uncertaintycited by 901
  6. 2023A Comprehensive Survey on Test-Time Adaptation under Distribution Shiftscited by 359
  7. 2021Open-Set Recognition: a Good Closed-Set Classifier is All You Need?cited by 353
  8. 2022VOS: Learning What You Don't Know by Virtual Outlier Synthesiscited by 300
  9. 2021SSD: A Unified Framework for Self-Supervised Outlier Detectioncited by 264
  10. 2022Extremely Simple Activation Shaping for Out-of-Distribution Detectioncited by 212
  11. 2023OpenOOD v1.5: Enhanced Benchmark for Out-of-Distribution Detectioncited by 179
  12. 2023Non-Parametric Outlier Synthesiscited by 129
8 more
  1. 2022How to Exploit Hyperspherical Embeddings for Out-of-Distribution Detection?cited by 128
  2. 2020BIMCV COVID-19+: a large annotated dataset of RX and CT images from COVID-19 patientscited by 82
  3. 2023Scaling for Training Time and Post-hoc Out-of-distribution Detection Enhancementcited by 77
  4. 2023ID-like Prompt Learning for Few-Shot Out-of-Distribution Detectioncited by 42
  5. 2024Envisioning Outlier Exposure by Large Language Models for Out-of-Distribution Detectioncited by 32
  6. 2023AUTO: Adaptive Outlier Optimization for Test-Time OOD Detectioncited by 13
  7. 2024CLIPScope: Enhancing Zero-Shot OOD Detection with Bayesian Scoringcited by 13
  8. 2023Likelihood-Aware Semantic Alignment for Full-Spectrum Out-of-Distribution Detectioncited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Dual Distribution Estimation for Zero-shot Noisy Test-Time Adaptation with VLMscited by 2
  2. 2026TINS: Test-time ID-prototype-separated Negative Semantics Learning for OOD Detectioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024AdaNeg: Adaptive Negative Proxy Guided OOD Detection with Vision-Language Models
  2. 2026Debiased Negative Mining Improves Out-of-distribution Detection with Pre-trained Vision-Language Models
  3. 2025ANTS: Adaptive Negative Textual Space Shaping for OOD Detection via Test-Time MLLM Understanding and Reasoning
  4. 2024Negative Label Guided OOD Detection with Pretrained Vision-Language Models
  5. 2024CMOOD: Concept-based Multi-label OOD Detection
  6. 2026Fi-ImageNet-1k: An OOD Benchmark From the Inside of the ImageNet-1k Validation Set
  7. 2025SimLabel: Consistency-Guided OOD Detection with Pretrained Vision-Language Models
  8. 2026TTL: Test-time Textual Learning for OOD Detection with Pretrained Vision-Language Models