DeepFashion-MultiModal

DeepFashion-multipal是一个大规模的高质量人类数据集，具有丰富的多模态注释。它具有以下属性:
它包含44,096高分辨率人体图像，包括12,701全身人体图像。
对于每个全身图像，我们手动注释24个类的人类解析标签。
对于每个全身图像，我们手动注释关键点。
我们为每个人类图像提取DensePose。
每个图像都手动注释了衣服形状和纹理的属性。
我们为每个图像提供文本描述。

DeepFashion-multipal is a large-scale, high-quality human dataset with rich multimodal annotations. It has the following attributes:
It contains 44,096 high-resolution human images, including 12,701 full-body human images.
For each full-body image, 24 categories of human parsing labels were manually annotated.
Keypoints were manually annotated for each full-body image.
DensePose was extracted for each human image.
Each image was manually annotated with attributes related to clothing shape and texture.
Text descriptions are provided for each image.

搜集汇总

背景与挑战

背景概述

DeepFashion-MultiModal是一个大规模高质量的多模态时尚数据集，包含44,096张高分辨率人体图像，其中12,701张全身图像，每张图像均提供人类解析、关键点、DensePose、衣服属性和文本描述等丰富注释。该数据集由南洋理工大学于2022年发布，适用于计算机视觉、时尚分析和多模态学习等研究任务。