FashionAI 服饰属性标签识别数据集

With the upgrading of consumption concepts, the fashion industry has huge market potential. According to statistics from authoritative organizations, the global market value of the fashion industry has exceeded US$3 trillion. At the same time, AI technology is also developing continuously, but the technology still faces many challenges in the process of integrating with the fashion industry.
In order to promote the combination of AI technology and clothing fashion, Alibaba's "Image Harmony" team teamed up with the Department of Textiles and Clothing of the Hong Kong Polytechnic University to launch the industry's first large-scale high-quality fashion data set that meets both clothing professionalism and machine learning requirements, focusing on machines. There are two basic issues in cognitive fashion: clothing key point positioning and clothing attribute label identification. The clothing attribute label recognition data set was generated under this background. Clothing attribute tags are an important foundation for the clothing knowledge system, which is huge and complex. We have professionally organized and abstracted the clothing attributes, and built a label knowledge system that is consistent with the cognitive process, structured and meets the requirements of machine learning. The clothing attribute tag recognition technology born from this can be widely used in clothing image retrieval, tag navigation, clothing matching and other application scenarios.
The image data is collected from Alibaba e-commerce data.
This research topic focuses on local attribute identification of clothing products. All clearly identifiable attribute labels in the picture require prediction. Considering the complexity of clothing knowledge, this data set only retains the product image data of a single subject (single model or single piece tile), so that researchers can focus on solving the challenges in the attribute labeling task.