DTD stands for Describable Textures Dataset. It is a continuously updated texture dataset consisting of 5,640 images, which are divided into 47 categories based on human perception. Each category has 120 images. For each image, a list of key attributes and joint attributes is also provided. The purpose of this dataset is to further understand how intelligent systems process, analyze and represent texture information; compared with classic texture recognition, DTD perception features are richer in structure and variety, and more challenging. Example image:

Citation

@InProceedings{cimpoi14describing, Author = {M. Cimpoi and S. Maji and I. Kokkinos and S. Mohamed and A. Vedaldi}, Title = {Describing Textures in the Wild}, Booktitle = {Proceedings of the {IEEE} Conf. on Computer Vision and Pattern Recognition ({CVPR})}, Year = {2014} }