Dataset containing around 29843 images of cats' faces of size 64x64.

The images were retrieved from 4 different open datasets, namely:

• Cats and Dogs Breeds Classification Oxford Dataset (https://www.kaggle.com/zippyz/cats-and-dogs-breeds-classification-oxford-dataset)
• Cute Cats and Dogs from Pixabay.com (https://www.kaggle.com/ppleskov/cute-cats-and-dogs-from-pixabaycom)
• Cat Dataset (https://www.kaggle.com/crawford/cat-dataset)
• COCO dataset (http://cocodataset.org/#home)
• Cats faces 64x64 (For generative models) (https://www.kaggle.com/spandan2/cats-faces-64x64-for-generative-models/downloads/cats-faces-64x64-for-generative-models.zip/1)

The images were preprocessed with an openCV cat face detector (https://www.pyimagesearch.com/2016/06/20/detecting-cats-in-images-with-opencv/), so that each image would contain only the face of the cats. The Cat Dataset has been preprocessed twice, so that it would contain also a piece of the background in the image. The preprocessing was skipped for the last dataset (Cats faces 64x64), since the faces were alrady cropped there. At the end, the false positive cats were removed by a visual inspection of the dataset.