Clo-Sket

Clo-Sket is a database of clothing sketch pictures that had been produced based on six categories of clothing photograph images, including dresses, blouses, shirts, pants, skirts, and jackets. These images were further decomposed into 23 sub-categories. Ten photo samplings from each sub-category were sampled, making up 230 photograph images. Based on each photograph, ten sketchers sketched the clothing, resulted in a total of 2300 clothing sketch images. The dataset can be used for cross-domain (between photo and sketch) image matching, image retrieval. It can be used for training, testing and validating the classification algorithm based on machine learning or deep learning, due to the amount of the sketch image and sketch’s intra variability within one sub-category.

Clo-Sket是一款基于六大类服装照片生成的服装素描图像数据库，涵盖连衣裙、女式衬衫、衬衫、长裤、半身裙与夹克六大品类。该数据集所依托的原始照片被进一步细分为23个子类别，从每个子类别中抽取10张照片，总计得到230张服装照片。针对每张照片，由10名绘图者分别绘制该服装的素描作品，最终累计生成2300张服装素描图像。该数据集可应用于跨域（照片与素描之间）图像匹配与图像检索任务。得益于单个子类别内素描图像的样本量充足，且同类素描存在类内变异性，该数据集还可用于训练、测试与验证基于机器学习或深度学习的分类算法。