README.md

metadata

license: openrail
tags:
  - fashion
  - clothing
  - virtual-try-on
  - e-commerce
  - flatlay
  - image-generation
pretty_name: Fashion 1K
size_categories:
  - 1K<n<10K
task_categories:
  - object-detection
language:
  - en

Fashion 1K

Dataset Summary

Fashion 1K is a curated collection of 1,000 high-quality fashion images, focusing on apparel and outfit compositions without human models.

Unlike typical street-style datasets (like DeepFashion) that include human poses and complex backgrounds, this dataset provides clean, human-free images. The images primarily feature Flat Lay (clothing arranged on a flat surface) or Ghost Mannequin styles, making them ideal for tasks that require a clear view of the garment's structure, texture, and color without occlusion.

Key Features:

• Human-Free: No faces, limbs, or skin tones—strictly focused on the garments.
• Outfit-Centric: Many images showcase complete looks (e.g., Top + Bottom + Shoes) to aid in compatibility learning.
• Clean Backgrounds: Minimized background noise to facilitate easier segmentation and feature extraction.

Supported Tasks

This dataset is particularly suitable for:

• Virtual Try-On (VTON): Serving as the "garment" reference image (g_img) for 2D try-on pipelines.
• Fashion Compatibility Learning: Learning which items (e.g., shirt and trousers) go well together based on the curated outfits.
• Generative AI Training: Training LoRAs or ControlNets for specific clothing styles without the bias of human figures.
• E-commerce Tagging: Automated classification of clothing categories and attributes.

Dataset Structure

Data Fields

• image (image): The high-resolution image of the clothing item or outfit.

Usage Example

from datasets import load_dataset

# Load the dataset
ds = load_dataset("Codatta/Fashion-1K", split="train")

# Display the first image
sample = ds[0]
sample['image'].show()