Bogazici-DeviantArt Image Reuse Database


	BODAIR

What is BODAIR?

Detecting when an image is used in another image automatically is a challenging problem. BODAIR is a new digital art database that consists of stock images and artworks that re-use these stock images. Artistic re-use is not constrained and does not have a system, which makes the dataset challenging. The images are collected from DeviantArt with their permission and collaboration. DeviantArt is a social network for artists and art enthusiasts with more than 38 million registered users. DeviantArt members post over 160,000 images every day. Images posted under the stock image category are usually published under an open license and are free to use by others.

Content of Database

Examples from six subcategories of stock images: animals, food, nature, places, plants and premade backgrounds.

Key features

Six subcategories of stock images: animals, food, nature, places, plants and premade backgrounds.
200 images in each subcategories.
144 stock(original) images.
1056 query images.
Manually annotated for four reuse types and nine manipulation types.

Examples from reuse and manipulations(from top-left to bottom-right): duplicative use, direct reuse, use as background, aspect ratio change, alpha blending, color manipulation, rotation, translation, texture manipulation, partial reuse, text overlay, remake and mirroring. See the paper for more details and reuse types.

How to Obtain for Research and Non-Commercial Use

To receieve BODAIR, first download, fill out and sign the agreement form. Please note that the signee of the agreement must hold a permanent position at an academic instution.
Via your institutional e-mail address, send an e-mail to salah[at]boun.edu.tr which includes your purpose of usage and the signed agreeement form.
Your request will be reviewed and the download details will be sent to your institutional e-mail address.

Paper and Supplementary Material

Isikdogan, F., Adiyaman, I., Akdag Salah, A.A. and Salah, A.A., 2016, October.
A New Database and Protocol for Image Reuse Detection.
European Conference on Computer Vision (pp. 903-916). Springer International Publishing.
(hosted on Springer) | (Uncorrected author proof)

[Bibtex]