spikingjelly.datasets.cifar10_dvs#

class spikingjelly.datasets.cifar10_dvs.CIFAR10DVS(root, data_type='event', frames_number=None, split_by=None, duration=None, custom_integrate_function=None, custom_integrated_frames_dir_name=None, transform=None, target_transform=None)[源代码]#

基类：NeuromorphicDatasetFolder

API Language - 中文 | English

  • 中文

CIFAR10-DVS 数据集，由 CIFAR10-DVS: An Event-Stream Dataset for Object Classification 提出。

有关参数的更多详细信息，请参考 NeuromorphicDatasetFolder

  • English

The CIFAR10-DVS dataset, which is proposed by CIFAR10-DVS: An Event-Stream Dataset for Object Classification.

Refer to NeuromorphicDatasetFolder for more details about params information.

参数:

  • root (Union[str, Path]) -- 数据集的根路径

  • data_type (str) -- "event" 或 "frame"

  • frames_number (Optional[int]) -- 积分帧的数量

  • split_by (Optional[str]) -- "time" 或 "number"

  • duration (Optional[int]) -- 每帧的时间时长

  • custom_integrate_function (Optional[Callable]) -- 用户自定义积分函数

  • custom_integrated_frames_dir_name (Optional[str]) -- 自定义积分帧目录名

  • transform (Optional[Callable]) -- 数据变换

  • target_transform (Optional[Callable]) -- 标签变换

  • root -- Root directory of the dataset

  • data_type -- "event" or "frame"

  • frames_number -- Number of frames to integrate

  • split_by -- "time" or "number"

  • duration -- Time duration per frame

  • custom_integrate_function -- User-defined integrate function

  • custom_integrated_frames_dir_name -- Custom frames directory name

  • transform -- Transform function

  • target_transform -- Target transform function

classmethod get_H_W()[源代码]#

API Language - 中文 | English

  • 中文

返回:

(128, 128)

返回类型:

Tuple

  • English

返回:

(128, 128)

返回类型:

Tuple

classmethod resource_url_md5()[源代码]#
返回类型:

list

classmethod downloadable()[源代码]#
返回:

True

返回类型:

bool

classmethod extract_downloaded_files(download_root, extract_root)[源代码]#
参数:

  • download_root (Path)

  • extract_root (Path)

classmethod create_raw_from_extracted(extract_root, raw_root)[源代码]#
参数:
class spikingjelly.datasets.cifar10_dvs.CIFAR10DVSTEBNSplit(root, train=True, data_type='event', frames_number=None, split_by=None, duration=None, custom_integrate_function=None, custom_integrated_frames_dir_name=None, transform=None, target_transform=None)[源代码]#

基类：CIFAR10DVS

API Language - 中文 | English

  • 中文

CIFAR10-DVS 数据集，由 CIFAR10-DVS: An Event-Stream Dataset for Object Classification 提出。

原始的 CIFAR10-DVS 数据集不提供训练集和测试集的划分。 在 Temporal Effective Batch Normalization in Spiking Neural Networks 中， 作者使用每个类别中的样本 0-99 作为测试集，100-999 作为训练集。 这种划分被后来的工作广泛使用。此类实现了这种划分。

备注

在此划分上的验证准确率通常远高于随机划分的准确率。进行比较时要小心！

有关参数的更多详细信息，请参考 NeuromorphicDatasetFolder

  • English

The CIFAR10-DVS dataset, which is proposed by CIFAR10-DVS: An Event-Stream Dataset for Object Classification.

The original CIFAR10-DVS dataset does not provide train and test split. In Temporal Effective Batch Normalization in Spiking Neural Networks , the authors use sample 0-99 in each class as the test set, and the 100-999 as the train set. This split is widely used by later works. This class implements this split.

备注

The validation accuracy on this split is typically much higher than that on a random split. Be careful when making comparisons!

Refer to NeuromorphicDatasetFolder for more details about params information.

参数:

  • root (Union[str, Path]) -- 数据集的根路径

  • train (bool) -- 是否使用训练集（True 或 False）

  • data_type (str) -- "event" 或 "frame"

  • frames_number (Optional[int]) -- 积分帧的数量

  • split_by (Optional[str]) -- "time" 或 "number"

  • duration (Optional[int]) -- 每帧的时间时长

  • custom_integrate_function (Optional[Callable]) -- 用户自定义积分函数

  • custom_integrated_frames_dir_name (Optional[str]) -- 自定义积分帧目录名

  • transform (Optional[Callable]) -- 数据变换

  • target_transform (Optional[Callable]) -- 标签变换

  • root -- Root directory of the dataset

  • train -- Whether to use training set (True) or test set (False)

  • data_type -- "event" or "frame"

  • frames_number -- Number of frames to integrate

  • split_by -- "time" or "number"

  • duration -- Time duration per frame

  • custom_integrate_function -- User-defined integrate function

  • custom_integrated_frames_dir_name -- Custom frames directory name

  • transform -- Transform function

  • target_transform -- Target transform function