Kaleidoscope: In-language Exams for Massively Multilingual Vision Evaluation
This addresses the problem of culturally inclusive evaluation for vision-language models, though it is incremental as it builds on existing multilingual benchmarks.
The authors tackled the lack of multilingual and multicultural evaluation for vision-language models by creating Kaleidoscope, a large-scale in-language multimodal benchmark covering 18 languages and 14 subjects with 20,911 questions, and found that top models perform poorly on low-resource languages and complex scenarios.
The evaluation of vision-language models (VLMs) has mainly relied on English-language benchmarks, leaving significant gaps in both multilingual and multicultural coverage. While multilingual benchmarks have expanded, both in size and languages, many rely on translations of English datasets, failing to capture cultural nuances. In this work, we propose Kaleidoscope, as the most comprehensive exam benchmark to date for the multilingual evaluation of vision-language models. Kaleidoscope is a large-scale, in-language multimodal benchmark designed to evaluate VLMs across diverse languages and visual inputs. Kaleidoscope covers 18 languages and 14 different subjects, amounting to a total of 20,911 multiple-choice questions. Built through an open science collaboration with a diverse group of researchers worldwide, Kaleidoscope ensures linguistic and cultural authenticity. We evaluate top-performing multilingual vision-language models and find that they perform poorly on low-resource languages and in complex multimodal scenarios. Our results highlight the need for progress on culturally inclusive multimodal evaluation frameworks.