What is Fashionpedia

Fashionpedia has several features:

  • 294 fine-grained apparel attributes
  • exhaustive high-quality segmentations masks for 27 main apparel categories, 19 apparel parts
  • a new task that combines both detection and attributes classification
  • expert-built ontology
  • trained models & API
  • kaggle challenges (iMat-Fashion)
  • Main contributors

  • Menglin Jia Cornell
  • Mengyun Shi Cornell
  • Mikhail Sirotenko Google
  • Yin Cui Google
  • Claire Cardie Cornell
  • Hartwig Adam Google
  • Bharath Hariharan Cornell
  • Van Dyk Lewis Cornell
  • Serge Belongie Cornell
  • Sponsors

    Image placeholder
    Image placeholder

    Citation

    									
    @inproceedings{jia2020fashionpedia,
      title={Fashionpedia: Ontology, Segmentation, and an Attribute Localization Dataset},
      author={Jia, Menglin and Shi, Mengyun and Sirotenko, Mikhail and Cui, Yin and Cardie, Claire and Hariharan, Bharath and Adam, Hartwig and Belongie, Serge}
      booktitle={European Conference on Computer Vision (ECCV)},
      year={2020}
    }
    	

    The Mission of Fashionpedia

    Visual analysis of clothing is a topic that has received increasing attention in recent years. Being able to recognize apparel products and associated attributes from pictures could enhance shopping experience for consumers, and increase work efficiency for fashion professionals.

    We present a new clothing dataset with the goal of introducing a novel fine-grained segmentation task by joining forces between the fashion and computer vision communities. The proposed task unifies both categorization and segmentation of rich and complete apparel attributes, an important step toward real-world applications.

    Fashionpedia is an ongoing effort that consists of:

    A novel task of fine-grained instance segmentation with attribute localization. The proposed task unifies instance segmentation and visual attribute recognition, which is an important step toward structural understanding of visual content in real-world applications.

    A unified fashion ontology informed by product descriptions from the internet and built by fashion experts. Our ontology captures the complex structure of fashion objects and ambiguity in descriptions obtained from the web, containing 46 apparel objects (27 main apparels and 19 apparel parts), and 294 fine-grained attributes (spanning 9 super categories) in total. To facilitate the development of related efforts, we also provide a mapping with categories from existing fashion segmentation datasets.

    A dataset with a total of 48,825 clothing images in daily-life, street-style, celebrity events, runway, and online shopping annotated both by crowd workers for segmentation masks and fashion experts for localized attributes, with the goal of developing and benchmarking computer vision models for comprehensive understanding of fashion.