imageInputLayer Normalization - Data normalization options

11 Ansichten (letzte 30 Tage)
Andrea Daou
Andrea Daou am 27 Aug. 2021
Kommentiert: Andrea Daou am 30 Sep. 2021
Hello,
In layer = imageInputLayer(inputSize,Name,Value), it is mentioned that the Normalization value can take 'zerocenter', 'zscore', 'rescale-zero-one', etc.
When choosing 'rescale-zero-one' as a 'Normalization' value I am getting an error telling me that only 'zerocenter' and 'none' can be taken as input for 'Normalization'. I checked imageInputLayer function and it is correct only these options exist (I am using MATLAB R2019a).
Do the other options exist in other MATLAB versions? Is it possible to use them?
Thank you in advance for your help!

Akzeptierte Antwort

Tomaso Cetto
Tomaso Cetto am 30 Sep. 2021
Bearbeitet: Tomaso Cetto am 30 Sep. 2021
Hi Andrea,
The 'rescale-zero-one' normalization option was added in R2019b, so you would need to update to at least that release to be able to make use of it!
Tomaso

Weitere Antworten (1)

yanqi liu
yanqi liu am 27 Sep. 2021
sir,may be you should update MATLAB version, such as 2021
inputlayer = imageInputLayer([28 28 3],'Name','input', 'Normalization', 'rescale-zero-one')
inputlayer =
ImageInputLayer with properties: Name: 'input' InputSize: [28 28 3] Hyperparameters DataAugmentation: 'none' Normalization: 'rescale-zero-one' NormalizationDimension: 'auto' Max: [] Min: []

Kategorien

Mehr zu Deep Learning Toolbox finden Sie in Help Center und File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by