HASY is a single-symbol dataset similar to the MNIST dataset. The dataset contains 168,233 instances covering 369 categories. The size of each image is 32 x 32 pixels. The dataset contains two challenges:

• A classification challenge with 10 predefined folders for 10-fold cross validation
• A verification challenge.

Citation

@dataset{thoma_2017_259444, author = {Thoma, Martin}, title = {HASYv2 – Handwritten Symbol database}, month = jan, year = 2017, publisher = {Zenodo}, doi = {10.5281/zenodo.259444}, url = {https://doi.org/10.5281/zenodo.259444}, }