---
license: mit
task_categories:
- zero-shot-classification
language:
- en
tags:
- fashion
- clothes
- pattern
size_categories:
- 10K<n<100K
---

This dataset consists of 10,898 images, which are composed of 19 different patterns.

The data was collected in a manner that ensures maximum balance across each pattern.
Minimum is 552(argyle) and maximum is 634(zebra).

## 1. Patterns

```
1. argyle
2. camouflage(military)
3. checked
4. dot
5. floral
6. geometric
7. gradient(gradation)
8. graphic
9. houndstooth
10. leopard
11. lettering
12. muji
13. paisley
14. snake skin
15. snow flake
16. stripe
17. tropical
18. zebra
19. zigzag
```

## 2. Product categories in images

Tried to include as wide a variety of product categories as possible to prevent the model biased. These categories include outerwear, tops, bottoms, hats, shoes, underwear, scarves, ties, socks, phone cases, and so on.


## 3. Data source

The sources of this data are Musinsa, SSF, Amazon, eBay, Pinterest, and Google Image Search.