regnet

RegNet Model API

This module provides utility functions to generate RegNet models.

Fore more information, please refer to the following paper: https://arxiv.org/abs/2101.00590

Parameters are in helia_edge.models.regnet_params.

Functions

Name	Description
build	RegNet model from RegNetParams
regnet_core	RegNet core
regnet_layer	Generate RegNet model

Machine-readable model

• yblockfunctionRegNet Y-Block
• zblockfunctionRegNet X-Block
• regnet_corefunctionRegNet core
• regnet_layerfunctionCreate RegNet TF functional model
• buildfunctionBuild a RegNet model.

function

yblock

Python

RegNet Y-Block

helia_edge/models/regnet.py:28

yblock(
    output_filters: int = 0,
    group_width: int = 0,
    kernel_size: int | tuple[int, int] = 3,
    strides: int | tuple[int, int] = 1,
    se_ratio: float = 4,
    activation: str = 'relu6',
    name: str | None = None,
) -> Callable[[keras.KerasTensor], keras.KerasTensor]

RegNet Y-Block

Parameters

Parameters of yblock

Name	Type	Default	Description
output_filters	int	0	Number of output filters. Defaults to 0.
group_width	int	0	Group width. Defaults to 0.
kernel_size	int | tuple[int, int]	3	Kernel size. Defaults to 3.
strides	int | tuple[int, int]	1	Stride length. Defaults to 1.
se_ratio	float	4	SE ratio. Defaults to 4.
activation	str	'relu6'	Activation function. Defaults to "relu6".
name	str | None	None	Block name. Defaults to None.

Returns

Returns of yblock

Type	Description
Callable[[keras.KerasTensor], keras.KerasTensor]	Callable[[keras.KerasTensor], keras.KerasTensor]: Functional layer

function

zblock

Python

RegNet X-Block

helia_edge/models/regnet.py:94

zblock(
    output_filters: int = 0,
    group_width: int = 0,
    kernel_size: int | tuple[int, int] = 3,
    strides: int | tuple[int, int] = 1,
    se_ratio: float = 4,
    activation: str = 'relu6',
    name: str | None = None,
) -> Callable[[keras.KerasTensor], keras.KerasTensor]

RegNet X-Block

Parameters

Parameters of zblock

Name	Type	Default	Description
output_filters	int	0	Number of output filters. Defaults to 0.
group_width	int	0	Group width. Defaults to 0.
kernel_size	int | tuple[int, int]	3	Kernel size. Defaults to 3.
strides	int | tuple[int, int]	1	Stride length. Defaults to 1.
se_ratio	float	4	SE ratio. Defaults to 4.
activation	str	'relu6'	Activation function. Defaults to "relu6".
name	str | None	None	Block name. Defaults to None.

Returns

Returns of zblock

Type	Description
Callable[[keras.KerasTensor], keras.KerasTensor]	Callable[[keras.KerasTensor], keras.KerasTensor]: Functional layer

function

regnet_core

Python

RegNet core

helia_edge/models/regnet.py:158

regnet_core(
    blocks: list[RegNetBlockParam],
    block_style: Literal['y', 'z'] = 'y',
) -> Callable[[keras.KerasTensor], keras.KerasTensor]

RegNet core

Parameters

Parameters of regnet_core

Name	Type	Default	Description
blocks	list[RegNetBlockParam]	Required	Block params
block_style	float	'y'	Block style. Defaults to 'y'.

Returns

Returns of regnet_core

Type	Description
Callable[[keras.KerasTensor], keras.KerasTensor]	Callable[[keras.KerasTensor], keras.KerasTensor]: Core

function

regnet_layer

Python

Create RegNet TF functional model

helia_edge/models/regnet.py:194

regnet_layer(x: keras.KerasTensor, params: RegNetParams) -> keras.KerasTensor

Create RegNet TF functional model

Parameters

Parameters of regnet_layer

Name	Type	Default	Description
x	keras.KerasTensor	Required	Input tensor
params	RegNetParams	Required	Model parameters.

Returns

Returns of regnet_layer

Type	Description
keras.KerasTensor	keras.KerasTensor: Output tensor

function

build

Python

Build a RegNet model.

helia_edge/models/regnet.py:248

build(
    params: RegNetParams,
    input_shape: tuple[int | None, ...],
    *,
    batch_size: int | None = None,
    name: str | None = None,
) -> keras.Model

Build a RegNet model.

Parameters

Parameters of build

Name	Type	Default	Description
params	RegNetParams	Required	Model parameters.
input_shape	tuple[int | None, ...]	Required	Input shape without the batch axis; None for a variable axis.
batch_size	int | None	None	Static batch size; None for a dynamic batch.
name	str | None	None	Model name; the family when None.

Returns

Returns of build

Type	Description
keras.Model	keras.Model: The model, named ``regnet`` unless ``name`` is given.