Fashion Dataset

Used for training machine learning models. It is compiled from a number of other datasets.

Fashionista Dataset (158,235 images)

Thanks to Kota Yamaguchi for the fashionista dataset. Paper

The fashionista package contains the Fashionista dataset without annotation, which was collected from chictopia.com in 2011.

The data is stored in a tab-delimited text files in the following format. Text files are split into chunks. Concatenating them will recover the full data records in one table.

posts/xxx.txt

Post(id, url)

Post represents a blog post in chictopia.com. In Chictopia, bloggers upload up to 5 photos to a single post. This table keeps the url of these posts and a unique identifier in the dataset.

Note that a blogger might delete some of the posts since the dataset was collected. It's not guaranteed that all posts are available.

photos/xxx.txt

Photo(post_id, url)

Photo represents a picture associated to each post. In the table, one row keeps a url and the id of the associated post.

garments/xxx.txt

Garment(post_id, name)

Garment is meta-data extracted from the post. In each post, bloggers list up their clothing items from the pre-defined set of clothing types. This garment table keeps pairs of the id of the post and extracted garment name.