FoodieQA

FoodieQA: A Multimodal Dataset for Fine-Grained Understanding of Chinese Food Culture

Github Repo

😋 We release all tools and code used to create the dataset at https://github.com/lyan62/FoodieQA.

Paper

For more details about the dataset, please refer to 📄 FoodieQA: A Multimodal Dataset for Fine-Grained Understanding of Chinese Food Culture

Dataset Download

!!Note!! The Json files are in the FoodieQA.zip (click on the Files and… See the full description on the dataset page: https://huggingface.co/datasets/lyan62/FoodieQA.

FoodieQA：一款针对中国美食文化细粒度理解的跨模态数据集

搜集汇总

背景与挑战

背景概述

FoodieQA是一个多模态基准数据集，专注于中国饮食文化的细粒度理解，包含多图像和单图像视觉问答（VQA）以及文本问答（TextQA）任务。数据集基于389个独特食物图像和350个食物条目构建，图像来自志愿者而非网络，以确保评估公平性，支持中英文双语，适用于评估视觉语言模型在文化细节理解上的能力。