Mastering Negation: Boosting Grounding Models via Grouped Opposition-Based Learning

2026-03-12

AI Summary

This study addresses the challenge that existing vision-language grounding models struggle to accurately interpret and localize complex instructions containing negation due to a scarcity of high-quality negative semantic data. To this end, the authors introduce D-Negation, the first dataset comprising paired positive–negative semantic annotations, and propose a Grouped Contrastive Learning framework. This framework organizes semantically opposing descriptions in a structured manner, incorporates a complementary semantic contrastive loss, and leverages parameter-efficient fine-tuning—updating fewer than 10% of model parameters—to substantially enhance the model’s comprehension of negation. Experimental results demonstrate consistent improvements of 4.4 mAP and 5.7 mAP on positive and negative semantic grounding tasks, respectively, significantly boosting both robustness and accuracy.

Technology Category

• Natural Language Processing: Language Grounding & Multi-modal NLP
• Computer Vision: Language and Vision
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Current vision-language detection and grounding models predominantly focus on prompts with positive semantics and often struggle to accurately interpret and ground complex expressions containing negative semantics. A key reason for this limitation is the lack of high-quality training data that explicitly captures discriminative negative samples and negation-aware language descriptions. To address this challenge, we introduce D-Negation, a new dataset that provides objects annotated with both positive and negative semantic descriptions. Building upon the observation that negation reasoning frequently appears in natural language, we further propose a grouped opposition-based learning framework that learns negation-aware representations from limited samples. Specifically, our method organizes opposing semantic descriptions from D-Negation into structured groups and formulates two complementary loss functions that encourage the model to reason about negation and semantic qualifiers. We integrate the proposed dataset and learning strategy into a state-of-the-art language-based grounding model. By fine-tuning fewer than 10 percent of the model parameters, our approach achieves improvements of up to 4.4 mAP and 5.7 mAP on positive and negative semantic evaluations, respectively. These results demonstrate that explicitly modeling negation semantics can substantially enhance the robustness and localization accuracy of vision-language grounding models.

Problem

• negation
• vision-language grounding
• negative semantics
• object grounding
• language-based detection

Innovation

• negation-aware learning
• opposition-based learning
• vision-language grounding
• negative semantics
• D-Negation dataset