mlsanityData to 7 Oct 2026

Not Just What's There: Enabling CLIP to Comprehend Negated Visual Descriptions Without Fine-tuning

arXiv 2602.21035 ↗First posted 24 Feb 2026cs.CV cs.MMcited by 5

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

Abstract (arXiv)

Vision-Language Models (VLMs) like CLIP struggle to understand negation, often embedding affirmatives and negatives similarly (e.g., matching "no dog" with dog images). Existing methods refine negation understanding via fine-tuning CLIP's text encoder, risking overfitting. In this work, we propose CLIPGlasses, a plug-and-play framework that enhances CLIP's ability to comprehend negated visual descriptions. CLIPGlasses adopts a dual-stage design: a Lens module disentangles negated semantics from text embeddings, and a Frame module predicts context-aware repulsion strength, which is integrated into a modified similarity computation to penalize alignment with negated semantics, thereby reducing false positive matches. Experiments show that CLIP equipped with CLIPGlasses achieves competitive in-domain performance and outperforms state-of-the-art methods in cross-domain generalization. Its superiority is especially evident under low-resource conditions, indicating stronger robustness across domains.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Placed in Prompt learning and adaptation of CLIP, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 29 of its 47 references that are in the map. First posted on arXiv 24 Feb 2026.

Builds on

29 of 47 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024Extract Free Dense Misalignment from CLIP24 Dec
  2. 2022CREPE: Can Vision-Language Foundation Models Reason Compositionally?13 Dec
  3. 2022When and why vision-language models behave like bags-of-words, and what to do about it?4 Oct

Other references · 26

  1. 2017Attention Is All You Needcited by 59,932
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  3. 2016Layer Normalizationcited by 5,579
  4. 2016Gaussian Error Linear Units (GELUs)cited by 3,431
  5. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  6. 2022Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionalitycited by 523
  7. 2023BiomedCLIP: a multimodal biomedical foundation model pretrained from fifteen million scientific image-text pairscited by 506
  8. 2023Alpha-CLIP: A CLIP Model Focusing on Wherever You Wantcited by 150
  9. 2024Revisiting Catastrophic Forgetting in Large Language Model Tuningcited by 93
  10. 2021Theoretical Foundations of t-SNE for Visualizing High-Dimensional Clustered Datacited by 55
  11. 2024StitchFusion: Weaving Any Visual Modalities to Enhance Multimodal Semantic Segmentationcited by 38
  12. 2021RpBERT: A Text-image Relation Propagation-based BERT Model for Multimodal NERcited by 37
14 more
  1. 2024SemTalk: Holistic Co-speech Motion Generation with Frame-level Semantic Emphasiscited by 32
  2. 2025Exploring Efficient Open-Vocabulary Segmentation in the Remote Sensingcited by 31
  3. 2025CLIP Behaves like a Bag-of-Words Model Cross-modally but not Uni-modallycited by 30
  4. 2025Know "No" Better: A Data-Driven Approach for Enhancing Negation Awareness in CLIPcited by 26
  5. 2023TagCLIP: A Local-to-Global Framework to Enhance Open-Vocabulary Multi-Label Classification of CLIP Without Trainingcited by 19
  6. 2025EchoMask: Speech-Queried Attention-based Mask Modeling for Holistic Co-Speech Motion Generationcited by 17
  7. 2024How and where does CLIP process negation?cited by 16
  8. 2025Recurrence-Enhanced Vision-and-Language Transformers for Robust Multimodal Document Retrievalcited by 13
  9. 2025MadaKV: Adaptive Modality-Perception KV Cache Eviction for Efficient Multimodal Long-Context Inferencecited by 12
  10. 2024CLIPVQA:Video Quality Assessment via CLIPcited by 11
  11. 2025Bringing CLIP to the Clinic: Dynamic Soft Labels and Negation-Aware Learning for Medical Analysiscited by 10
  12. 2024FedCFA: Alleviating Simpson's Paradox in Model Aggregation with Counterfactual Federated Learningcited by 4
  13. 2025SPJFNet: Self-Mining Prior-Guided Joint Frequency Enhancement for Ultra-Efficient Dark Image Restorationcited by 3
  14. 2025Beyond Illumination: Fine-Grained Detail Preservation in Extreme Dark Image Restorationcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026CreativeBench: Benchmarking and Enhancing Machine Creativity via Self-Evolving Challengescited by 9
  2. 2026NEGATE: Constrained Semantic Guidance for Linguistic Negation in Text-to-Video Diffusioncited by 2
  3. 2026DCR: Counterfactual Attractor Guidance for Rare Compositional Generationcited by 1
  4. 2026TopoAgent: An Agentic Framework for Automated Topology Learning in Medical Imagingcited by 0
  5. 2026Beyond Visual Similarity: Entity-Aligned Retrieval for Knowledge-Based Visual Question Answeringcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026When Negation Is a Geometry Problem in Vision-Language Models
  2. 2024Learn "No" to Say "Yes" Better: Improving Vision-Language Models via Negations
  3. 2026Omni-NegCLIP: Enhancing CLIP with Front-Layer Contrastive Fine-Tuning for Comprehensive Negation Understanding
  4. 2026What CLIP Knows but Cannot Say: Recovering Negation from Frozen Intermediate Features
  5. 2025TNG-CLIP:Training-Time Negation Data Generation for Negation Awareness of CLIP
  6. 2025VITRIX-CLIPIN: Enhancing Fine-Grained Visual Understanding in CLIP via Instruction Editing Data and Long Captions
  7. 2026HANCLIP: A Family of Hyperbolic Angular Negation Vision Language Models
  8. 2022CLIP-TD: CLIP Targeted Distillation for Vision-Language Tasks