unext

U-NeXt

Overview

U-NeXt is a modification of U-Net that utilizes techniques from ResNeXt and EfficientNetV2. During the encoding phase, mbconv blocks are used to efficiently process the input.

Parameters are in helia_edge.models.unext_params.

Functions

Name	Description
build	U-NeXt model from UNextParams
unext_block	Create U-NeXt block
se_block	Squeeze and excite block
norm_layer	Normalization layer
unext_core	Create U-NeXt core
unext_layer	Create U-NeXt layer

Additions

The U-NeXt architecture has been modified to allow the following:

• MBConv blocks used in the encoding phase.
• Squeeze and excitation (SE) blocks added within blocks.

Machine-readable model

• se_blockfunctionSqueeze and excite block
• norm_layerfunctionNormalization layer
• unext_blockfunctionCreate UNext block
• unext_corefunctionCreate UNext TF functional core
• unext_layerfunctionCreate UNext TF functional model
• buildfunctionBuild a UNext model.

function

se_block

Python

Squeeze and excite block

helia_edge/models/unext.py:35

se_block(ratio: int = 8, name: str | None = None)

Squeeze and excite block

function

norm_layer

Python

Normalization layer

helia_edge/models/unext.py:61

norm_layer(norm: str, name: str) -> keras.Layer

Normalization layer

Parameters

Parameters of norm_layer

Name	Type	Default	Description
norm	str	Required	Normalization type
name	str	Required	Name

Returns

Returns of norm_layer

Type	Description
keras.Layer	keras.Layer: Layer

function

unext_block

Python

Create UNext block

helia_edge/models/unext.py:91

unext_block(
    output_filters: int,
    expand_ratio: float = 1,
    kernel_size: int | tuple[int, int] = 3,
    strides: int | tuple[int, int] = 1,
    se_ratio: float = 4,
    dropout: float | None = 0,
    norm: Literal['batch', 'layer'] | None = 'batch',
    name: str | None = None,
) -> keras.Layer

Create UNext block

function

unext_core

Python

Create UNext TF functional core

helia_edge/models/unext.py:181

unext_core(x: keras.KerasTensor, params: UNextParams) -> keras.KerasTensor

Create UNext TF functional core

Parameters

Parameters of unext_core

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

Returns

Returns of unext_core

Type	Description
keras.KerasTensor	keras.KerasTensor: Output tensor

function

unext_layer

Python

Create UNext TF functional model

helia_edge/models/unext.py:328

unext_layer(inputs: keras.KerasTensor, params: UNextParams) -> keras.KerasTensor

Create UNext TF functional model

Parameters

Parameters of unext_layer

Name	Type	Default	Description
inputs	keras.KerasTensor	Required	Input tensor
params	UNextParams	Required	Model parameters.

Returns

Returns of unext_layer

Type	Description
keras.KerasTensor	keras.KerasTensor: Output tensor

function

build

Python

Build a UNext model.

helia_edge/models/unext.py:371

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

Build a UNext model.

Parameters

Parameters of build

Name	Type	Default	Description
params	UNextParams	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 ``unext`` unless ``name`` is given.