STMNIST#

class tonic.prototype.datasets.STMNIST(root: PathLike, keep_compressed: Optional[bool] = False, skip_sha256_check: Optional[bool] = True, shuffle: bool = False)[source]#

ST-MNIST

Neuromorphic Spiking Tactile MNIST (ST-MNIST) dataset, which comprises handwritten digits obtained by human participants writing on a neuromorphic tactile sensor array. The original paper can be found at https://arxiv.org/abs/2005.04319. Data is provided with the MAT format. Download of the compressed dataset has to be done by the user by accessing https://scho larbank.nus.edu.sg/bitstream/10635/168106/2/STMNIST%20dataset%20NUS%20Tee%20Research%20Group.zi p, where a form has to be completed. Then, the path to the ZIP archive has to be provided to the STMNIST constructor root argument.

Events have (xytp) ordering. :param root: Parent folder of ‘STMNIST/STMNIST dataset NUS Tee Research Group.zip’. The STMNIST folder is related to the Tonic class name and is needed currently. :type root: string :param shuffle: Whether to shuffle the dataset. More efficient if done based on file paths. :type shuffle: bool

Returns:

Torchdata data pipe that yields a tuple of events (or transformed events) and target.

Return type:

dp (IterDataPipe[Sample])

Parameters:

  • root (PathLike) –

  • keep_compressed (Optional[bool]) –

  • skip_sha256_check (Optional[bool]) –

  • shuffle (bool) –