SewFactory

该数据集名为SewFactory，包含了大约100万张图像及其对应的真实缝纫图案，用于模型训练和定量评估。它涵盖了广泛的人体姿态、体型以及缝纫图案。此外，SewFactory具有高姿态变异性，包含了多种衣物和人体纹理，有效缩小了与真实世界输入的领域差距。该数据集规模约为100万张图像，其任务是仅从单张图像中恢复缝纫图案。

The dataset, named SewFactory, comprises approximately 1 million images paired with their corresponding ground-truth sewing patterns, intended for model training and quantitative evaluation. It covers a wide range of human poses, body shapes, and sewing patterns. Furthermore, SewFactory features high pose variability and incorporates diverse clothing and human textures, effectively narrowing the domain gap between the dataset and real-world inputs. The task of this dataset is to recover sewing patterns solely from a single image.

搜集汇总

背景与挑战

背景概述

SewFactory是一个合成数据集，包含约1M图像和真实缝制图案，用于从单张图像恢复服装缝制图案。该数据集具有高姿势变异性和多样化的服装及人体纹理，有效缩小了与真实世界输入的领域差距，并提供了丰富的真实标签，适用于多种应用场景。