WORLD WIDE DISHES

WORLD WIDE DISHES数据集由牛津人工智能协会等多个国际研究机构合作创建，包含765种来自世界各地的食品数据，涵盖131种本地语言的食品名称。数据集通过建立一个可通过社交媒体广泛分发的网站，采用去中心化的方式收集，旨在通过食品这一文化代理，评估基础模型在不同地区的性能和代表性偏差。数据集的应用领域包括旅行者的食品识别系统、文化研究等，旨在解决基础模型在全球范围内表现不均的问题。

The WORLD WIDE DISHES dataset was collaboratively created by the Oxford Artificial Intelligence Association and several other international research institutions. It encompasses 765 food-related datasets originating from across the globe, covering food names in 131 local languages. The dataset was collected in a decentralized manner by building a website that enables wide dissemination via social media. Its core objective is to assess the performance and representational biases of foundation models across different regions, using food as a cultural proxy. Its application areas include food recognition systems for travelers, cultural research, and other related fields, aiming to address the issue of uneven performance of foundation models on a global scale.