Datasets:
metadata
pretty_name: The Culture Caption Dataset
language:
- en
license: cc-by-4.0
tags:
- fashion
- captions
- multimodal
task_categories:
- image-to-text
size_categories:
- n<1K
image-captioner2
Dataset Summary
- Total labelled looks: 16
- Total images: 73
- Size bucket (examples): n<1K
- Archetype labels: athleisure, casual, classic, formal, minimalist, preppy, streetwear
- Detected items covered: belt, dress, long sleeved hoodie, long sleeved shirt, pullover, scarf, short sleeved shirt, tie, trousers, vest
- Colour palette: beige, black, brown, dark blue, gray, grey, lemongrass, light blue, navy, red, taupe-beige, white
- Materials represented: cotton, cotton blend, denim, elanstane, fleece, leather, metal, polyester, rayon, satin, wool, wool blend
- Style tags observed: athleisure, boxy, button up, casual, chinos, classic, corset, cuffed, drop shoulder, fitted, formal, graphic, half-zip, jeans, kangaroo pocket, light wash, mini dress, minimalist, oversized, palazzo, pleated, polo, preppy, quarter-zip, regular fit, ribbed, rolled-up sleeves, scrunch-bottom, slacks, slim fit, solid, strapless, streetwear, striped, suit, sweatpants, thumbholes, wide leg
Usage
Each record links an image with caption metadata, detected items, colours, materials, style tags, and an archetype label.
Load captions_base.json directly for the full annotations, or stream the images folder for fine-tuning pipelines.
Update Log
Latest upload via automation on 2026-01-02 03:01 UTC.