HALC: Object Hallucination Reduction via Adaptive Focal-Contrast Decoding

Zhaorun Chen, Zhuokai Zhao, Hongyin Luo, Huaxiu Yao, Bo Li, Jiawei Zhou

Keywords: Object Hallucination, Trustworthy Vision-Language Models, Multimodal Foundation Models, Image-to-text Generation, Adaptive Decoding

TL;DR: Based on observations with optimal visual context, we propose a novel adaptive decoding method to address the object hallucination issues in image-to-text generation tasks for vision-language models.

Abstract: While large vision-language models (LVLMs) have demonstrated impressive capabilities in interpreting multi-modal contexts, they invariably suffer from object hallucinations (OH). We introduce HALC, a novel decoding algorithm designed to mitigate OH in LVLMs. HALC leverages distinct fine-grained optimal visual information in vision-language tasks and operates on both local and global contexts simultaneously. Specifically, HALC integrates a robust auto-focal grounding mechanism (locally) to correct hallucinated tokens on the fly, and a specialized beam search algorithm (globally) to significantly reduce OH while preserving text generation quality. Additionally, HALC can be integrated into any LVLMs as a plug-and-play module without extra training. Extensive experimental studies demonstrate HALC’s effectiveness in reducing OH, outperforming state-of-the-arts across four benchmarks. Code is released at https://github.com/BillChan226/HALC.

Submission Number: 62