Back to datasets
Dataset assetOpen Source CommunityMachine LearningImage Processing

yuukicammy/MIT-Adobe-FiveK

The MIT‑Adobe FiveK dataset is a publicly available collection containing 5,000 RAW images in DNG format, each retouched by five experts to produce 25,000 TIFF images (16‑bit per channel, ProPhoto RGB, lossless). The dataset also includes semantic information for each image. Created by MIT and Adobe Systems, Inc., it is intended to provide a diverse and challenging test set for image‑processing algorithms. Images cover a wide range of scenes—landscapes, portraits, still life, architecture—and exhibit varied lighting, color balance, and exposure conditions.

Source
hugging_face
Created
Nov 28, 2025
Updated
Apr 16, 2023
Signals
918 views
Availability
Linked source ready
Overview

Dataset description and usage context

Dataset Overview

Name: MIT‑Adobe FiveK Dataset

Task Category:

  • image‑to‑image

Tags:

  • RAW
  • raw
  • DNG
  • dng
  • denoising
  • superresolution
  • underexposure
  • overexposure

Friendly Name: fiveK

Size Category:

  • 1K < n < 10K

Dataset Content

  1. Raw Images: 5,000 RAW files in DNG format.
  2. Expert‑Retouched Images: Each RAW image was edited by five experts, yielding 25,000 TIFF files with the following properties:
    • 16‑bit per channel
    • ProPhoto RGB colour space
    • Lossless compression
  3. Semantic Information: Metadata for each image, including shooting location, time, lighting conditions, and subject.

Intended Use

The dataset, created by MIT and Adobe Systems, Inc., serves as a diverse and challenging benchmark for testing image‑processing algorithms. It includes scenes such as landscapes, portraits, still‑life, and architecture, with variations in illumination, colour balance, and exposure.

Dataset Structure

The dataset has a complex hierarchy; this repository provides tools for convenient download and usage. Each sample comprises the original DNG image, five expert‑edited versions, detailed semantic metadata, and camera model information.

Example Data

Raw (DNG)Expert AExpert BExpert CExpert DExpert ECategoriesCamera Model
a0001‑jmac_DSC1459.dngtiff16_a/a0001‑jmac_DSC1459tiff16_b/a0001‑jmac_DSC1459tiff16_c/a0001‑jmac_DSC1459tiff16_d/a0001‑jmac_DSC1459tiff16_e/a0001‑jmac_DSC1459{"location":"outdoor","time":"day","light":"sun_sky","subject":"nature"}Nikon D70
a1384‑dvf_095.dngtiff16_a/a1384‑dvf_095tiff16_b/a1384‑dvf_095tiff16_c/a1384‑dvf_095tiff16_d/a1384‑dvf_095tiff16_e/a1384‑dvf_095{"location":"outdoor","time":"day","light":"sun_sky","subject":"nature"}Leica M8
a4607‑050801_080948__I2E5512.dngtiff16_a/a4607‑050801_080948__I2E5512tiff16_b/a4607‑050801_080948__I2E5512tiff16_c/a4607‑050801_080948__I2E5512tiff16_d/a4607‑050801_080948__I2E5512tiff16_e/a4607‑050801_080948__I2E5512{"location":"indoor","time":"day","light":"artificial","subject":"people"}Canon EOS‑1D Mark II
Need downstream help?

Pair the dataset with AI analysis and content workflows.

Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.

Explore AI studio