spikingjelly.datasets package#

备注

API 改动说明

SpikingJelly 0.0.0.1.0 对 datasets 包进行了结构性重构，以提升代码的模块化程度和可维护性。主要改动为：

  • datasets/__init__.py 被拆分为 base.py、 utils.py 以及 transform.py ；

  • 新的 datasets/__init__.py 仅作为 API 门面（facade），负责引入并暴露各子模块中的公开对象；

  • datasets/to_rep_x.py 及其中内容已被移除。

使用建议：

  • 数据集类：推荐从 spikingjelly.datasets 命名空间直接导入；

  • 工具函数和数据增强：推荐通过 spikingjelly.datasets.utils 和 spikingjelly.datasets.transform 子模块访问；

# 推荐 ✅
from spikingjelly.datasets import DVS128Gesture
from spikingjelly.datasets.utils import create_sub_dataset

# 旧的导入方式，不推荐 ❌
from spikingjelly.datasets.dvs128_gesture import DVS128Gesture
from spikingjelly.datasets import create_sub_dataset

为保证兼容性，以上旧的导入方式在当前版本中仍然可用，但不再推荐在新代码中使用。

API Change Notice

In SpikingJelly 0.0.0.1.0, the datasets package has undergone a structural refactor to improve modularity and maintainability. The main changes include:

  • datasets/__init__.py has been split into base.py, utils.py, and transform.py;

  • the new datasets/__init__.py now serves solely as an API facade, responsible for importing and re-exporting public objects from its submodules;

  • datasets/to_rep_x.py and its contents have been removed.

Usage recommendations:

  • Dataset classes: import them directly from the spikingjelly.datasets namespace;

  • Utility functions and data augmentation: access them via the spikingjelly.datasets.utils and spikingjelly.datasets.transform submodules;

# Recommended ✅
from spikingjelly.datasets import DVS128Gesture
from spikingjelly.datasets.utils import create_sub_dataset

# Legacy import patterns, not recommended ❌
from spikingjelly.datasets.dvs128_gesture import DVS128Gesture
from spikingjelly.datasets import create_sub_dataset

To preserve compatibility, the legacy import patterns shown above remain supported in the current version, but their use is discouraged.

Datasets#

Dataset Base Class#

Dataset Utilities#

Data Augmentations#