The DogFaceNet datasets

The DogFaceNet Datasets

Datasets for dog face detection and identification.

This record contains three datasets:

DogFaceNet_large is the raw dog face recognition dataset. It contains 2,522 folders of images, each folder containing images of one dog. Disclaimer: This dataset is not fully curated and some mistakes are present (e.g., black and white images instead of colored ones, multiple individual dogs in the same folder, etc.).

DogFaceNet_alignment is the dog face detection dataset. It contains one folder of images and one CSV file. The CSV file has seven columns: filename, (x, y) coordinates of the left eye, the right eye, and the nose of each dog.

DogFaceNet_224resized is the aligned and resized dog face recognition dataset. It contains 1,393 folders of images. Each image is a JPG of size (224, 224, 3) with a single aligned dog face. The classes_test.txt and classes_train.txt files list which dog classes were used to train and test the deep learning models.

For any questions or issues about the datasets, please open an issue on the GitHub page: https://github.com/GuillaumeMougeot/DogFaceNet.