README.md

metadata

license: odc-by
task_categories:
  - image-classification
  - image-to-text
  - text-to-image
language:
  - en
  - fr
  - zh
  - ja
  - ko
size_categories:
  - 100K<n<1M
tags:
  - fashion
  - clothing
dataset_info:
  features:
    - name: 1.jpg
      dtype: image
    - name: 2.jpg
      dtype: image
    - name: 3.jpg
      dtype: image
    - name: 4.jpg
      dtype: image
    - name: json
      struct:
        - name: categories
          sequence:
            dtype: string
            length: -1
        - name: designer
          struct:
            - name: en
              dtype: string
            - name: fr
              dtype: string
            - name: ja
              dtype: string
            - name: ko
              dtype: string
            - name: zh
              dtype: string
        - name: gender
          dtype: string
        - name: id
          dtype: string
        - name: name
          struct:
            - name: en
              dtype: string
            - name: fr
              dtype: string
            - name: ja
              dtype: string
            - name: ko
              dtype: string
            - name: zh
              dtype: string
        - name: price
          dtype: string
        - name: url
          dtype: string

Fashion-150k

Product images + metadata

Stats:

• 151,748 products
• 587,824 images
• 1,186 designers

Formats:

• WebDataset
  • 1k per shard, ~0.4GB
  • N.jpg + json
  • __key__ is id from json

Fields

• gender: men/women
• name: includes translations for en, fr, zh, ja and ko
• categories: main category and sub-categories e.g. ['CLOTHING', 'TOPS', 'T-SHIRTS'], ['BAGS', 'TOTE BAGS']
• designer: includes translations for en, fr, zh, ja and ko
• price: in USD

another BIG data banger straight from the underground

with thanks to the retailer and all designers ❤️