Mitigating Object and Action Hallucinations in Multimodal LLMs via Self-Augmented Contrastive Alignment

TL;DR

This paper introduces SANTA, a framework that reduces hallucinations in multimodal language models by identifying false correlations and aligning visual facts with captions, improving factual accuracy in video descriptions.

Contribution

The paper presents a novel self-augmented contrastive alignment method specifically designed to mitigate object and action hallucinations in dynamic video captioning models.

Findings

SANTA significantly reduces hallucinations in generated video captions.

The method outperforms existing approaches on hallucination benchmarks.

Enhanced alignment improves factual consistency in multimodal descriptions.

Abstract

Recent advancement in multimodal LLMs (MLLMs) has demonstrated their remarkable capability to generate descriptive captions for input videos. However, these models suffer from factual inaccuracies in the generated descriptions, causing severe hallucination issues. While prior works have explored alleviating hallucinations for static images, jointly mitigating visual object and temporal action hallucinations for dynamic videos remains a challenging and unsolved task. To tackle this challenge, we propose a Self-Augmented Contrastive Alignment (SANTA) framework for enabling object and action faithfulness by exempting the spurious correlations and enforcing the emphasis on visual facts. SANTA employs a hallucinative self-augmentation scheme to identify the potential hallucinations that lie in the MLLM and transform the original captions to the contrasted negatives. Furthermore, we develop a tracklet-phrase contrastive alignment to match the regional objects and relation-guided actions with their corresponding visual and temporal phrases. Extensive experiments demonstrate that SANTA outperforms existing methods in alleviating object and action hallucinations, yielding superior performance on the hallucination examination benchmarks.