Inference Preprocessing

Summary Table

Name	Interface	Access	Visibility	Description
Training Pixel Format	IEnumeration	RW	Beginner	The bit depth and number of input channels of the images used during training of the network that is on the camera. This is used when doing image normalization of the captured images before performing inference.
Network Input Channel	IEnumeration	RW	Beginner	Input channel of loaded network.
Training Mean	IFloat	RW	Beginner	Mean value of the data used during network training. This is used when doing image normalization of the captured images before performing inference. If the selected network type has only one input channel then the channel 1 mean is used.
Training Scale	IFloat	RW	Beginner	Standard deviation of the data used during network training. This is used when doing image normalization of the captured images before performing inference. If the selected network type has only one input channel then the channel 1 mean is used.

Inference Preprocessing Features

Training Pixel Format

The bit depth and number of input channels of the images used during training of the network that is on the camera. This is used when doing image normalization of the captured images before performing inference.

Property	Value
Name	InferencePreprocessBppSelector
Interface	IEnumeration
Access	RW
Visibility	Beginner

Enumeration Values
Mono8
RGB8
BGR8

Network Input Channel

Input channel of loaded network.

Property	Value
Name	InferencePreprocessChannelSelector
Interface	IEnumeration
Access	RW
Visibility	Beginner

Enumeration Values
Channel1
Channel2
Channel3

Training Mean

Mean value of the data used during network training. This is used when doing image normalization of the captured images before performing inference. If the selected network type has only one input channel then the channel 1 mean is used.

Property	Value
Name	InferencePreprocessChannelMean
Interface	IFloat
Access	RW
Unit
Visibility	Beginner

Training Scale

Standard deviation of the data used during network training. This is used when doing image normalization of the captured images before performing inference. If the selected network type has only one input channel then the channel 1 mean is used.

Property	Value
Name	InferencePreprocessChannelScalar
Interface	IFloat
Access	RW
Unit
Visibility	Beginner