Overview
The Wild-Anim dataset available from this page consists of 5 classes that contains uniformly distributed images examples of wild animals. In total, the dataset contains 5000 images. The images are slightly anamorphic in their representation. These images were fed as input to both deep learning and classical computer vision methods for carrying out animal recognition task. In our research, we considered two subsets each containing 1000 different examples from the total image examples. The dataset and the used subsets with details of the data-split and meta-descriptions of the class labels can be downloaded (here).

Publication

• "E. Okafor, P. Pawara, F. Karaaba, O. Surinta, V. Codreanu, L. Schomaker and M. Wiering " Comparative Study Between Deep Learning and Bag of Visual Words for Wild-Animal Recognition , in IEEE Symposium Series on Computational Intelligence (SSCI) 2016, at Athens in Greece."

Citation

@inproceedings{okafor2016comparative,
  title={Comparative study between deep learning and bag of visual words for wild-animal recognition},
  author={Okafor, Emmanuel and Pawara, Pornntiwa and Karaaba, Faik and Surinta, Olarik and Codreanu, Valeriu and Schomaker, Lambert and Wiering, Marco},
  booktitle={Computational Intelligence (SSCI), 2016 IEEE Symposium Series on},
  pages={1--8},
  year={2016},
  organization={IEEE}
}