Saved in:
Bibliographic Details
Main Authors: Suzuki, Teppei, Ozawa, Keisuke
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.09979
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908317776871424
author Suzuki, Teppei
Ozawa, Keisuke
author_facet Suzuki, Teppei
Ozawa, Keisuke
contents We propose an efficient evaluation protocol for large vision-language models (VLMs). Given their broad knowledge and reasoning capabilities, multiple benchmarks are needed for comprehensive assessment, making evaluation computationally expensive. To improve efficiency, we construct a subset that yields results comparable to full benchmark evaluations. Our benchmark classification experiments reveal that no single benchmark fully covers all challenges. We then introduce a subset construction method using farthest point sampling (FPS). Our experiments show that FPS-based benchmarks maintain a strong correlation (> 0.96) with full evaluations while using only ~1\% of the data. Additionally, applying FPS to an existing benchmark improves correlation with overall evaluation results, suggesting its potential to reduce unintended dataset biases.
format Preprint
id arxiv_https___arxiv_org_abs_2504_09979
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Resampling Benchmark for Efficient Comprehensive Evaluation of Large Vision-Language Models
Suzuki, Teppei
Ozawa, Keisuke
Computer Vision and Pattern Recognition
We propose an efficient evaluation protocol for large vision-language models (VLMs). Given their broad knowledge and reasoning capabilities, multiple benchmarks are needed for comprehensive assessment, making evaluation computationally expensive. To improve efficiency, we construct a subset that yields results comparable to full benchmark evaluations. Our benchmark classification experiments reveal that no single benchmark fully covers all challenges. We then introduce a subset construction method using farthest point sampling (FPS). Our experiments show that FPS-based benchmarks maintain a strong correlation (> 0.96) with full evaluations while using only ~1\% of the data. Additionally, applying FPS to an existing benchmark improves correlation with overall evaluation results, suggesting its potential to reduce unintended dataset biases.
title Resampling Benchmark for Efficient Comprehensive Evaluation of Large Vision-Language Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.09979