Learning from Fine-Grained Visual Discrepancies: Mitigating Multimodal Hallucinations via In-Context Visual Contrastive Optimization
About
Multimodal hallucination remains a persistent challenge for Vision-Language Models (VLMs). Standard textual Direct Preference Optimization (DPO) often fails to mitigate it due to a lack of explicit visual supervision. While existing works introduce visual preference DPO by contrasting original images against negative ones, they suffer from a theoretically inconsistent objective caused by partition function mismatches and rely on coarse-grained negatives that could enable shortcut learning. In this work, we propose In-Context Visual Contrastive Optimization (IC-VCO). By placing contrastive images within a shared multi-image context, IC-VCO ensures a mathematically rigorous objective. We further introduce Visual Contrast Distillation (VCDist), an auxiliary reliability-gated regularizer that encourages consistency between multi-image contrastive training and single-image inference. Finally, we propose a contrastive sample editing strategy that generates hard negatives via precise semantic perturbations. Experiments on five benchmarks demonstrate IC-VCO's best overall performance and the effectiveness of our sample editing strategy. Code and data are available at https://github.com/OPPO-Mente-Lab/IC-VCO.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Visual Perception | BLINK | -- | 241 | |
| Hallucination Evaluation | HallusionBench | -- | 153 | |
| Visual Hallucination Evaluation | HallusionBench | -- | 120 | |
| Visual Perception and Reasoning | BLINK | Accuracy49.01 | 64 | |
| Multi-modal Hallucination Evaluation | AMBER | -- | 28 | |
| Compositional Reasoning and Perception Evaluation | CRPE | Exist Score94.15 | 13 | |
| General Multimodal Evaluation | Macro-average of HallusionBench, AMBER, CRPE, R-Bench, and BLINK | Overall Score63.35 | 13 | |
| Multimodal Hallucination Evaluation | CRPE | Existence Score96.8 | 13 | |
| Multimodal Hallucination Evaluation | R-Bench | Dis66.68 | 13 | |
| Robustness | R-Bench | R-Bench Dis Metric60.71 | 13 |