sensorutils.core

`to_frames(src, window_size, stride, stride_mode='index')`

np.ndarrayをフレーム分けするプリミティブな実装で，stride_modeで分割アルゴリズムを指定することが可能．

to_frames関数はto_frames_using_index関数，to_frames_using_nptricks関数，to_frames_using_reshape関数を適応的に使い分ける．

使い分けは以下の通り．

window_size == stride -> to_frames_using_reshape
window_size != stride and stride_mode == 'index' -> to_frames_using_index
window_size != stride and stride_mode == 'nptrick' -> to_frames_using_nptricks

Parameters:

Name	Type	Description	Default
`src`	`ndarray`	splited source.	required
`window_size`	`int`	sliding window size.	required
`stride`	`int`	stride is int more than 0.	required
`stride_mode`	`str`	'index' or 'nptrick'. it is used `to_frames_*` method when window_size != stride.	`'index'`

Returns:

Type	Description
`ndarray`	a shape of frames is `(num_frames, window_size, *src.shape[1:])`, where num_frames is `(src.shape[0] - window_size) // stride + 1`.

`to_frames_using_reshape(src, window_size)`

np.ndarrayをフレーム分けするプリミティブな実装で，ウィンドウサイズとストライド幅が同じ場合に利用することが可能．

分割にnp.reshapeを使用しており，非常に高速なsliding-window処理を実行可能．

Parameters:

Name	Type	Description	Default
`src`	`ndarray`	splited source.	required
`window_size`	`int`	sliding window size. stride = window_size.	required

Returns:

Type	Description
`ndarray`	a shape of frames is `(num_frames, window_size, *src.shape[1:])`, where num_frames is `(src.shape[0] - window_size) // window_size + 1`.

`to_frames_using_index(src, window_size, stride)`

np.ndarray をフレーム分けするプリミティブな実装で，分割にindexingを使用している．

Parameters:

Name	Type	Description	Default
`src`	`ndarray`	splited source.	required
`window_size`	`int`	sliding window size.	required
`stride`	`int`	stride is int more than 0.	required

Returns:

Type	Description
`ndarray`	a shape of frames is `(num_frames, window_size, *src.shape[1:])`, where num_frames is `(src.shape[0] - window_size) // stride + 1`.

`to_frames_using_nptricks(src, window_size, stride)`

np.ndarray をフレーム分けするプリミティブな実装で，分割にnp.lib.stride_tricks.as_strided関数を使用しており，indexingを使用するto_frames_using_indexより高速である．

Parameters:

Name	Type	Description	Default
`src`	`ndarray`	splited source.	required
`window_size`	`int`	sliding window size.	required
`stride`	`int`	stride is int more than 0.	required

Returns:

Type	Description
`ndarray`	a shape of frames is `(num_frames, window_size, *src.shape[1:])`, where num_frames is `(src.shape[0] - window_size) // stride + 1`.

`split_using_sliding_window(segment, **options)`

可変サイズのsegmentからsliding-window方式で一定サイズのフレームを抽出する．

各shapeは以下のようになる．

segment: (segment_size, ch)
frames: (num_frames, window_size, ch)

segmentの第2軸(axis=1)以降のshapeは任意であり，例えばshapeが(segment_size, ch1, ch2)のデータをsegmentとして入力すると， (num_frames, window_size, ch1, ch2)のデータを取得することができる．

Parameters:

Name	Type	Description	Default
`segments`	`np.ndarray`	分割対象のデータ	required
`window_size`	`int, default=512`	フレーム分けするサンプルサイズ	required
`stride`	`int, default=None`	strideがNoneはwindow_sizeが指定される．	required
`ftrim`	`int`	最初のftrimサンプルをとばす(default=5)．	required
`btrim`	`int`	最後のbtrimサンプルをとばす(default=5)．	required
`return_error_value`	`None`	エラーの時の返り値	required

Returns:

Type Description

ndarray

sliding-window方式で抽出下フレーム

失敗したらreturn_error_valueで指定された値を返す．

`split_using_target(src, target)`

targetのデータを元にsrcの分割を行う．

Parameters:

Name	Type	Description	Default
`src`	`ndarray`	分割するデータ	required
`target`	`ndarray`	ラベルデータ(一次元配列)	required

Examples:

>>> tgt = np.array([0, 0, 1, 1, 2, 2, 1])
>>> src = np.array([1, 2, 3, 4, 5, 6, 7])
>>> splited = split_from_target(src, tgt)
>>>
>>> # splited == {
>>> #    0: [np.array([1, 2])],
>>> #    1: [np.array([3, 4]), np.array([7])],
>>> #    2: [np.array([5, 6])]
>>> # }

Returns:

Type	Description
`Dict[int, List[numpy.ndarray]]`	keyはラベル，valueはデータのリスト．

`interpolate(src, rate, kind='linear', axis=-1)`

interpolation function.

Use scipy.interpolate.interp1d in this function.

Parameters:

Name	Type	Description	Default
`src`	`ndarray`	interpolation source.	required
`rate`	`int`	rate.	required
`kind`	`str`	Specifies the kind of interpolation as a string (‘linear’, ‘nearest’, ‘zero’, ‘slinear’, ‘quadratic’, ‘cubic’, ‘previous’, ‘next’, where ‘zero’, ‘slinear’, ‘quadratic’ and ‘cubic’ refer to a spline interpolation of zeroth, first, second or third order; ‘previous’ and ‘next’ simply return the previous or next value of the point) or as an integer specifying the order of the spline interpolator to use. Default is ‘linear’.	`'linear'`
`axis`	`int`	Specifies the axis of y along which to interpolate. Interpolation defaults to the last axis of y.	`-1`

Examples:

Linear interpolation

1 2	`[0, 2, 4] - x2 -> [0, 1, 2, 3, 4] [0, 3, 6] - x3 -> [0, 1, 2, 3, 4, 5, 6]`

Returns:

Type	Description
`ndarray`	shape[axis] - 1 == rate * ft.shape[axis]

`pickle_dump(obj, path)`

object dump using pickle.

Parameters:

Name	Type	Description	Default
`obj`	`Any`	any object.	required
`path`	`Union[str, pathlib.Path]`	save path.	required

`pickle_load(path)`

object load using pickle.

Parameters:

Name	Type	Description	Default
`path`	`Path`	a saved object pickle path.	required