Seeing Clearly without Training: Mitigating Hallucinations in Multimodal LLMs for Remote Sensing

Yi Liu, Jing Zhang, Di Wang, Xiaoyu Tian, Haonan Guo +1 more
3/3/2026
cs.CV

Official code: MiliLab/RADAR (11 stars, last commit 2026-03-05).

Abstract

Multimodal large language models (MLLMs) suffer from pronounced hallucinations in remote sensing visual question-answering (RS-VQA), primarily caused by visual grounding failures in large-scale scenes or misinterpretation of fine-grained small targets. To systematically analyze these issues, we introduce RSHBench, a protocol-based benchmark for fine-grained diagnosis of factual and logical hallucinations. To mitigate grounding-induced factual hallucinations, we further propose Relative Attention-Driven Actively Reasoning (RADAR), a training-free inference method that leverages intrinsic attention in MLLMs to guide progressive localization and fine-grained local reasoning at test time. Extensive experiments across diverse MLLMs demonstrate that RADAR consistently improves RS-VQA performance and reduces both factual and logical hallucinations. Code and data will be publicly available at: https://github.com/MiliLab/RADAR

View on arXivView PDF

Code Implementations(1)

MiliLab/RADAROfficial100%

Official repo for "Seeing Clearly without Training: Mitigating Hallucinations in Multimodal LLMs for Remote Sensing"

111Mar 1, 20267 months ago

Similar Papers

RSHallu: Dual-Mode Hallucination Evaluation for Remote-Sensing Multimodal Large Language Models with Domain-Tailored Mitigation

90% match2026

Cracking the Code of Hallucination in LVLMs with Vision-aware Head Divergence

86% match1 repo212024

Seeing is Believing: Rich-Context Hallucination Detection for MLLMs via Backward Visual Grounding

85% match1 repo102025

FiVL: A Framework for Improved Vision-Language Alignment through the Lens of Training, Evaluation and Explainability

85% match2024

Visual Description Grounding Reduces Hallucinations and Boosts Reasoning in LVLMs

84% match1 repo252026

Toward More Reliable Artificial Intelligence: Reducing Hallucinations in Vision-Language Models

84% match2025

Cite this paper

@article{liu2026seeing,
  title  = {Seeing Clearly without Training: Mitigating Hallucinations in Multimodal LLMs for Remote Sensing},
  author = {Yi Liu and Jing Zhang and Di Wang and Xiaoyu Tian and Haonan Guo and Bo Du},
  year   = {2026},
  eprint = {2603.02754},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2603.02754v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2603.02754.svg)](https://codeofpaper.com/paper/2603.02754)

Discussion