RePOPE: Impact of Annotation Errors on the POPE Benchmark

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Neuhaus, Yannic, Hein, Matthias
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910916409294848
author Neuhaus, Yannic
Hein, Matthias
author_facet Neuhaus, Yannic
Hein, Matthias
contents Since data annotation is costly, benchmark datasets often incorporate labels from established image datasets. In this work, we assess the impact of label errors in MSCOCO on the frequently used object hallucination benchmark POPE. We re-annotate the benchmark images and identify an imbalance in annotation errors across different subsets. Evaluating multiple models on the revised labels, which we denote as RePOPE, we observe notable shifts in model rankings, highlighting the impact of label quality. Code and data are available at https://github.com/YanNeu/RePOPE .
format Preprint
id arxiv_https___arxiv_org_abs_2504_15707
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RePOPE: Impact of Annotation Errors on the POPE Benchmark
Neuhaus, Yannic
Hein, Matthias
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Since data annotation is costly, benchmark datasets often incorporate labels from established image datasets. In this work, we assess the impact of label errors in MSCOCO on the frequently used object hallucination benchmark POPE. We re-annotate the benchmark images and identify an imbalance in annotation errors across different subsets. Evaluating multiple models on the revised labels, which we denote as RePOPE, we observe notable shifts in model rankings, highlighting the impact of label quality. Code and data are available at https://github.com/YanNeu/RePOPE .
title RePOPE: Impact of Annotation Errors on the POPE Benchmark
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2504.15707