line-corporation/JIC-VQA

JIC-VQA（日语图像分类视觉问答）数据集是一个评估日语视觉语言模型（VLMs）的基准。该数据集通过在每个样本中添加问题，基于recruit-jp的日本图像分类评估数据集构建而成。所有问题均为多项选择题，每个问题有四个选项，这些选项与相应标签紧密相关，以提高任务的难度。数据集中的图像涵盖了101种日本食品、30种日本花卉、20种日本设施和10种日本地标。

Japanese Image Classification Visual Question Answering (JIC-VQA) is a benchmark for evaluating Japanese Vision-Language Models (VLMs). It is built based on the recruit-jp/japanese-image-classification-evaluation-dataset by adding questions to each sample. All questions are multiple-choice with four options each, designed to increase the tasks difficulty by closely relating the options to their respective labels. The dataset includes images of 101 types of Japanese food, 30 types of Japanese flowers, 20 types of Japanese facilities, and 10 types of Japanese landmarks.