WorldCuisines: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines
This addresses the problem of cultural bias in AI for researchers and developers working on multilingual and multicultural applications, though it is incremental as it focuses on benchmarking rather than novel model improvements.
The authors tackled the problem of Vision Language Models (VLMs) struggling with culture-specific knowledge in non-English languages and underrepresented contexts by introducing WorldCuisines, a massive-scale multilingual and multicultural visual question answering benchmark with over 1 million data points across 30 languages, showing that VLMs perform better with correct location context but struggle with adversarial contexts and predicting specific regional cuisines and languages.
Vision Language Models (VLMs) often struggle with culture-specific knowledge, particularly in languages other than English and in underrepresented cultural contexts. To evaluate their understanding of such knowledge, we introduce WorldCuisines, a massive-scale benchmark for multilingual and multicultural, visually grounded language understanding. This benchmark includes a visual question answering (VQA) dataset with text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark to date. It includes tasks for identifying dish names and their origins. We provide evaluation datasets in two sizes (12k and 60k instances) alongside a training dataset (1 million instances). Our findings show that while VLMs perform better with correct location context, they struggle with adversarial contexts and predicting specific regional cuisines and languages. To support future research, we release a knowledge base with annotated food entries and images along with the VQA data.