Skip to content

Dataset · ToneStyle

Image triplets for color style transfer

ToneStyle/TST100K

108,808 sets of source photos, references, and results for training automated color correction algorithms.

What's inside

Contains 108,808 image triplets, where each example includes an original photograph, a reference photo, and a target retouched frame. The data is compiled from open datasets PPR10K, MIT-Adobe FiveK, and Food-101. The dataset's task is to reproduce the photographic color and tone of the reference shot on the source frame.

What you can do with it

  • Training color and tone style transfer models
  • Creating neural networks for automated color correction and photo retouching
  • Benchmarking reference-based color processing algorithms
Size
100K–1M rows
Downloads this month
37.2K
Likes
5

How to load it

from datasets import load_dataset

ds = load_dataset("ToneStyle/TST100K")