mbconv

MBConv Layer API

This module provides classes to build mobile inverted bottleneck convolutional layers.

Classes

Name	Description
MBConvParams	MBConv parameters

Functions

Name	Description
mbconv_block	MBConv block w/ expansion and SE

Machine-readable model

• mbconv_blockfunctionMBConv block w/ expansion and SE

function

mbconv_block

Python

MBConv block w/ expansion and SE

helia_edge/layers/mbconv.py:25

mbconv_block(
    output_filters: int,
    expand_ratio: float = 1,
    kernel_size: int | tuple[int, int] = 3,
    strides: int | tuple[int, int] = 1,
    se_ratio: float = 8,
    droprate: float = 0,
    bn_momentum: float = 0.9,
    activation: str | Callable = 'relu6',
    name: str | None = None,
) -> keras.Layer

MBConv block w/ expansion and SE

This layer can support 1D inputs by providing a dummy dimension. In such cases, the kernel_size and strides should be adjusted accordingly.

Parameters

Parameters of mbconv_block

Name	Type	Default	Description
output_filters	int	Required	Number of output filter channels
expand_ratio	float	1	Expansion ratio. Defaults to 1.
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	8	SE ratio. Defaults to 8.
droprate	float	0	Drop rate. Defaults to 0.
bn_momentum	float	0.9	Batch normalization momentum. Defaults to 0.9.
activation	str	'relu6'	Activation function. Defaults to "relu6".
name	str | None	None	Block name. Defaults to None.

Returns

Returns of mbconv_block

Type	Description
keras.Layer	keras.Layer: Functional layer