u_net: Creates a U-Net architecture.

Description Usage Arguments Value

Description

Creates a U-Net architecture.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
u_net(
  net_h,
  net_w,
  grayscale,
  blocks = 4,
  n_class = 2,
  filters = 16,
  dropout = 0.1,
  batch_normalization = TRUE,
  kernel_initializer = "he_normal"
)

Arguments

net_h

Input layer height. Must be equal to '2^x, x - natural'..

net_w

Input layer width. Must be equal to '2^x, x - natural'.

grayscale

Defines input layer color channels - '1' if 'TRUE', '3' if 'FALSE'.

blocks

Number of blocks in the model.

n_class

Number of classes. Minimum is '2' (background + other object).

filters

Integer, the dimensionality of the output space (i.e. the number of output filters in the convolution).

dropout

Dropout rate.

batch_normalization

Should batch normalization be used in the block.

kernel_initializer

Initializer for the kernel weights matrix.

Value

U-Net model.


maju116/platypus documentation built on Oct. 18, 2020, 9:40 a.m.