Reefknot: A Comprehensive Benchmark for Relation Hallucination Evaluation, Analysis and Mitigation in Multimodal Large Language Models
This work addresses relation hallucinations in MLLMs, a critical issue for improving trustworthiness in multimodal AI applications, though it is incremental as it builds on existing hallucination research.
The paper tackles the problem of relation hallucinations in multimodal large language models, which are more complex than object or attribute hallucinations and lack comprehensive benchmarks. It introduces Reefknot, a benchmark with over 20,000 samples, and proposes a confidence-based mitigation strategy that reduces hallucination rates by an average of 9.75% across datasets.
Hallucination issues continue to affect multimodal large language models (MLLMs), with existing research mainly addressing object-level or attribute-level hallucinations, neglecting the more complex relation hallucinations that require advanced reasoning. Current benchmarks for relation hallucinations lack detailed evaluation and effective mitigation, and their datasets often suffer from biases due to systematic annotation processes. To address these challenges, we introduce Reefknot, a comprehensive benchmark targeting relation hallucinations, comprising over 20,000 real-world samples. We provide a systematic definition of relation hallucinations, integrating perceptive and cognitive perspectives, and construct a relation-based corpus using the Visual Genome scene graph dataset. Our comparative evaluation reveals significant limitations in current MLLMs' ability to handle relation hallucinations. Additionally, we propose a novel confidence-based mitigation strategy, which reduces the hallucination rate by an average of 9.75% across three datasets, including Reefknot. Our work offers valuable insights for achieving trustworthy multimodal intelligence.