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

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yang, Zesheng, Jiang, Xi, Hu, Bingzhang, Guan, Weili, Cong, Runmin, Qi, Guo-Jun, Zheng, Feng
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911511886168064
author Yang, Zesheng
Jiang, Xi
Hu, Bingzhang
Guan, Weili
Cong, Runmin
Qi, Guo-Jun
Zheng, Feng
author_facet Yang, Zesheng
Jiang, Xi
Hu, Bingzhang
Guan, Weili
Cong, Runmin
Qi, Guo-Jun
Zheng, Feng
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2603_12606
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Mastering Negation: Boosting Grounding Models via Grouped Opposition-Based Learning
Yang, Zesheng
Jiang, Xi
Hu, Bingzhang
Guan, Weili
Cong, Runmin
Qi, Guo-Jun
Zheng, Feng
Computer Vision and Pattern Recognition
Artificial Intelligence
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.
title Mastering Negation: Boosting Grounding Models via Grouped Opposition-Based Learning
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2603.12606