SADI: Spatial-Aware Dynamic Intervention for Mitigating Hallucinations in LVLMs
Open Access
- 6 November 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 5498-5509
Abstract
No abstract availableFunding Information
- Shandong Provincial Natural Science Foundation (ZR2022MF333)
- Shandong Provincial Natural Science Foundation (ZR2024QF053)
- The Pilot Project for Integrated Innovation of Science, Education, and Industry of Qilu University of Technology (Shandong Academy of Sciences) (2026ZDCX01)
- The projects of the Key Laboratory of Computing Power Network and Information Security, Ministry of Education (2023ZD028)
This publication has 28 references indexed in Scilit:
- ICT: Image-Object Cross-Level Trusted Intervention for Mitigating Object Hallucination in Large Vision-Language Models2025
- HalLoc: Token-level Localization of Hallucinations for Vision Language Models2025
- Devils in Middle Layers of Large Vision-Language Models: Interpreting, Detecting and Mitigating Object Hallucinations via Attention Lens2025
- Mitigating Hallucinations in Large Vision-Language Models via DPO: On-Policy Data Hold the Key2025
- Seeing Far and Clearly: Mitigating Hallucinations in MLLMs with Attention Causal DecodingConference on Computer Vision and Pattern Recognition (CVPR), 2025
- Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding2024
- Improved Baselines with Visual Instruction Tuning2024
- OPERA: Alleviating Hallucination in Multi-Modal Large Language Models via Over-Trust Penalty and Retrospection-Allocation2024
- RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-Grained Correctional Human Feedback2024
- Microsoft COCO: Common Objects in Context2014