Release of Human-Rights-UNderstanding-CNNs

HRUN is the first ever image dataset of human rights violations photographs, labelled with human rights semantic categories, comprising a list of the types of human rights abuses encountered at present.

Data of Human Rights UNderstanding

Here we release the data of Human Rights UNderstanding to the public. HRUN has been used to train various CNNs based on a deep representation learning tool for image classification with Matlab, called MatDeepRep.

(HRUN) dataset consists of 4 different categories of human rights violations and 100 diverse images per category. This dataset was constructed with the specific purpose of capturing a range of real world situations of common human rights violations (e.g. child labour, child soldiers etc.).

Images for all 4 violation categories can be directly accessed here [download].

The per-category images can be downloaded here: [child labour] [child soldiers] [police violence] [refugees]

CNNs for image classification trained on HRUN:

For more information about trained CNNs on HRUN please refer to MatDeepRep repo.

Reference

Please cite the following paper if you use the data.

 @conference{visapp17,
 author={Grigorios Kalliatakis and Shoaib Ehsan and Maria Fasli and Ales Leonardis and Juergen Gall and Klaus D. McDonald-Maier},
 title={Detection of Human Rights Violations in Images: Can Convolutional Neural Networks Help?},
 booktitle={Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP, (VISIGRAPP 2017)},
 year={2017},
 pages={289-296},
 publisher={SciTePress},
 organization={INSTICC},
}

---

Please email Grigorios Kalliatakis if you have any questions or comments.