five

Visual Similarity Scores between Countries and Topics

收藏
arXiv2025-09-30 收录
下载链接:
https://github.com/MichiganNLP/visual_diversity_budget
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集包含了基于图像平均视觉表现计算的国家间主题视觉相似度得分。它旨在识别需要在视觉-语言模型中加强代表性的国家,以便进行标注,从而提升模型的表现。此外,数据集还包括了不同国家和主题的相似度得分,以及地理距离与视觉相似度之间的相关性洞察。任务目标是提高模型性能,通过识别需要标注的代表性不足的国家和主题。

This dataset contains cross-national thematic visual similarity scores calculated based on the average visual performance of images. It aims to identify countries that need enhanced representation in vision-language models for annotation, so as to improve model performance. Additionally, the dataset includes similarity scores for various countries and themes, as well as insights into the correlation between geographic distance and visual similarity. The core task objective is to enhance model performance by identifying underrepresented countries and themes that require annotation.
提供机构:
Michigan NLP
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作