AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

RFI: Rectified Flow Intervention for Mitigating Object Hallucination in Large Vision-Language Models

Junyu Cheng, Zhibiao Liang, Yidong Chen, Shuangyin Li

AAAI 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Preference tuning against hallucination · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes Rectified Flow Intervention (RFI), a method to mitigate object hallucination in LVLMs, evaluated from an abstract-only view.

Problem

Object hallucination in large vision-language models, where generated text is inconsistent with the input image; existing dynamic mitigation approaches have limitations (the abstract is truncated, so the specific limitations are not fully stated).

Approach

Uses rectified flow to intervene on LVLM hidden representations to steer generation toward image-consistent content; exact mechanism not stated in the available abstract text.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add RFI rectified-flow interventionLLaVA-v1.5 · POPE COCO randomF1: 81.33 -> 90.40Improved
add RFI rectified-flow interventionLLaVA-v1.5 · POPE A-OKVQA adversarialF1: 75.15 -> 82.00Improved
add RFI rectified-flow interventionLLaVA-v1.5 and Qwen-VL · POPE (average over subsets)F1: 7.59% (LLaVA-v1.5) and 6.90% (Qwen-VL)Improved
replace static ICT vector with dynamic RFIICT on LLaVA-v1.5 · POPE A-OKVQA randomF1: 89.41 -> 91.15Improved
replace static ICT vector with dynamic RFIICT on Qwen-VL · POPE COCO randomF1: 89.20 -> 88.89Hurt
replace ICT with RFIICT on Qwen-VL · POPE COCO adversarialF1: 84.42 -> 84.49No clear effect
replace ICT with RFIICT on LLaVA-v1.5 and Qwen-VL · MME hallucination subsetscore: 27.91 points (LLaVA-v1.5) and 47.48 points (Qwen-VL)Improved
replace ICT with RFIICT · MME countscoreHurt
move intervention layer away from layer 15 (ablation)RFI · POPE COCO randomF1Hurt
tune intensity alpha to about 5.0 (LLaVA) and 2.0 (Qwen-VL) (ablation)RFI · POPE COCO randomF1Improved
add RFI at inference (100 new tokens)LLaVA-v1.5 · Latencyseconds: 2.79 -> 3.05Hurt
add RFI at inference (5 new tokens)LLaVA-v1.5 · Latencyseconds: 0.28 -> 0.53Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy