DataScienceProject/Art_Images_Ai_And_Real_

Dataset Card for Dataset Name

This dataset card aims to give a basic understanding what is the data contain, where it come from and how it is structured. the purpose of such dataset exist.

Dataset Details

The dataset contains of 2,839 images: training set --> 2560 , test set -> 279.

The data is balanced in each set , 50% is label 0 (as real image) --> 50% label 1 (Ai generated image)

important :

The dataset is already ordered by folders so you can dowload/clone the repository and use the csv files to assing the labels (0,1) or do it with code.

Dataset Description

Art type : many art styles such as oil , water , drawings ect.

Images of art from Ai generators and Images of art human painters

Dataset Sources [optional]

AI generated images from the engines :

• Ideogram -https://ideogram.ai/login

• Civitai -https://civitai.com/

• Copilot

• Hugging Face:- stable-diffusion-xl-base-1.0

• openart.ai - https://openart.ai/

Real images :

WikiArt -https://www.wikiart.org/

The Met -https://www.metmuseum.org/art/collection

Unsplash -https://unsplash.com/

google image search

google art - https://artsandculture.google.com/partner/the-metropolitan-museum-of-art

art museum/gallery sites

Uses

This dataset created for a model that can distinguish between real and Ai images of art.

Direct Use

In this days when Ai image generation become a powerful tool it is very hard to tell if an image is real or fake (Ai generated), especially when we taking about an oil painting art

So we wanted to create a model that can classify images of art in order to solve a potential froud prevention that someone wants to sell an image of art claming he painted it digitally and the truth is that this is Ai generated image

Out-of-Scope Use

This dataset has of images contain art from famous painters like Vincent Van Gogh and this images used only for model traning.

The Ai generated images has no copyright on so you can use them as you like.

Dataset Structure

*** ela_output folder needed only for cnn+ela Model we created!