Aesthetic Visual Analysis

这是一个用于图像美学评估的数据集，包含250,000多个具有丰富元数据的图像。数据集共有66种类型的语义注释和14种类型的摄影风格注释，每幅图像都有美学得分。与现有数据集相比，该数据集在规模，注释的多样性和异质性方面具有优势，可用于研究审美偏好的计算模型。

This is a dataset for image aesthetic assessment, which contains more than 250,000 images with rich metadata. The dataset provides 66 types of semantic annotations and 14 types of photographic style annotations, and each image is attached with an aesthetic score. Compared with existing datasets, this dataset boasts advantages in terms of scale, annotation diversity and annotation heterogeneity, and can be employed to research computational models of aesthetic preferences.

搜集汇总

背景与挑战

背景概述

Aesthetic Visual Analysis是一个用于图像美学评估的大规模数据集，包含250,000多个图像，具有丰富的美学得分和多种注释类型，适用于审美偏好计算模型的研究。