# GLSim: Detecting Object Hallucinations in LVLMs via Global-Local Similarity

Source: [GLSim: Detecting Object Hallucinations in LVLMs via Global-Local Similarity](<https://www.sota2.com/research/paper/68fa7c1e1e209131cea8c7a8>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T05:56:14.357Z

Authors: Seongheon Park, Sharon Li

Published: 2025-08-27T15:30:06.000Z

## Abstract

Object hallucination in large vision-language models presents a significant challenge to their safe deployment in real-world applications. Recent works have proposed object-level hallucination scores to estimate the likelihood of object hallucination; however, these methods typically adopt either a global or local perspective in isolation, which may limit detection reliability. In this paper, we introduce GLSim, a novel training-free object hallucination detection framework that leverages complementary global and local embedding similarity signals between image and text modalities, enabling more accurate and reliable hallucination detection in diverse scenarios. We comprehensively benchmark existing object hallucination detection methods and demonstrate that GLSim achieves superior detection performance, outperforming competitive baselines by a significant margin.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Object Hallucination Detection on MSCOCO](<https://www.sota2.com/research/sota/object-hallucination-detection-on-mscoco.md>): Object Hallucination Detection; MSCOCO; AUROC: 85.36.

- [Object Hallucination Detection on Objects365](<https://www.sota2.com/research/sota/object-hallucination-detection-on-objects365.md>): Object Hallucination Detection; Objects365; AUROC: 72.16.

- [Object Hallucination Detection on CLEVR](<https://www.sota2.com/research/sota/object-hallucination-detection-on-clevr.md>): Object Hallucination Detection; CLEVR; AUROC: 70.26.

- [Object Hallucination Detection on POPE (average over three sampling strategies)](<https://www.sota2.com/research/sota/object-hallucination-detection-on-pope-average-over-three-sampling-strategies.md>): Object Hallucination Detection; POPE average over three sampling strategies; AUROC: 70.13.

- [Object Detection on COCO 2017 (val)](<https://www.sota2.com/research/sota/object-detection-on-coco-2017-val.md>): Object Detection; COCO 2017 (val); AP: 27.58.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2508.19972>)

[PDF](<https://arxiv.org/pdf/2508.19972.pdf>)