import os
from typing import Callable, Optional
import numpy as np
from tonic.dataset import Dataset
[docs]class DVSGesture(Dataset):
"""`IBM DVS Gestures <http://research.ibm.com/dvsgesture/>`_
.. note:: This is (exceptionally) a preprocessed version of the original dataset,
where recordings that originally contained multiple labels have already
been cut into respective samples. Also temporal precision is reduced to ms.
::
@inproceedings{amir2017low,
title={A low power, fully event-based gesture recognition system},
author={Amir, Arnon and Taba, Brian and Berg, David and Melano, Timothy and McKinstry, Jeffrey and Di Nolfo, Carmelo and Nayak, Tapan and Andreopoulos, Alexander and Garreau, Guillaume and Mendoza, Marcela and others},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
pages={7243--7252},
year={2017}
}
Parameters:
save_to (string): Location to save files to on disk.
train (bool): If True, uses training subset, otherwise testing subset.
transform (callable, optional): A callable of transforms to apply to the data.
target_transform (callable, optional): A callable of transforms to apply to the targets/labels.
transforms (callable, optional): A callable of transforms that is applied to both data and
labels at the same time.
"""
test_url = "https://figshare.com/ndownloader/files/38020584"
train_url = "https://figshare.com/ndownloader/files/38022171"
test_md5 = "56070e45dadaa85fff82e0fbfbc06de5"
train_md5 = "3a8f0d4120a166bac7591f77409cb105"
test_filename = "ibmGestureTest.tar.gz"
train_filename = "ibmGestureTrain.tar.gz"
classes = [
"Hand clapping",
"Right hand wave",
"Left hand wave",
"Right arm cw", # clockwise
"Right arm ccw", # counter-clockwise
"Left arm cw",
"Left arm ccw",
"Arm roll",
"Air drums",
"Air guitar",
"Other gestures",
]
sensor_size = (128, 128, 2)
dtype = np.dtype([("x", np.int16), ("y", np.int16), ("p", bool), ("t", np.int64)])
ordering = dtype.names
def __init__(
self,
save_to: str,
train: bool = True,
transform: Optional[Callable] = None,
target_transform: Optional[Callable] = None,
transforms: Optional[Callable] = None,
):
super().__init__(
save_to,
transform=transform,
target_transform=target_transform,
transforms=transforms,
)
self.train = train
if train:
self.url = self.train_url
self.file_md5 = self.train_md5
self.filename = self.train_filename
self.folder_name = "ibmGestureTrain"
else:
self.url = self.test_url
self.file_md5 = self.test_md5
self.filename = self.test_filename
self.folder_name = "ibmGestureTest"
if not self._check_exists():
self.download()
file_path = os.path.join(self.location_on_system, self.folder_name)
for path, dirs, files in os.walk(file_path):
dirs.sort()
for file in files:
if file.endswith("npy"):
self.data.append(path + "/" + file)
self.targets.append(int(file[:-4]))
[docs] def __getitem__(self, index):
"""
Returns:
a tuple of (events, target) where target is the index of the target class.
"""
events = np.load(self.data[index])
events[:, 3] *= 1000 # convert from ms to us
events = np.lib.recfunctions.unstructured_to_structured(events, self.dtype)
target = self.targets[index]
if self.transform is not None:
events = self.transform(events)
if self.target_transform is not None:
target = self.target_transform(target)
if self.transforms is not None:
events, target = self.transforms(events, target)
return events, target
[docs] def __len__(self):
return len(self.data)
def _check_exists(self):
return (
self._is_file_present()
and self._folder_contains_at_least_n_files_of_type(100, ".npy")
)
