Input image data directly from Variable and not DataStore?

Question

Travis Meyer 2022-10-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1815720-input-image-data-directly-from-variable-and-not-datastore

回答： Travis Meyer 2022-10-2

I've got neural data that I'm using to create the 'images' for a CNN model. I have a 4 dimensional array where the first 3 dimensions are the X,Y,RGB values, and the 4th is the image index (ex 1:107,000). Can I port this directly into Matlab's CNN models? So far I have to use the data to create PNGs in test/train folders, then import using the example from 'Create simple Deep Learning Network...' provided by Matlab's documentation, but it would be so much more efficient to pipe the data directly.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Travis Meyer 2022-10-2

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1815720-input-image-data-directly-from-variable-and-not-datastore#answer_1064885

Got it!!! It's actually easier than expected, I can pass the 4D array directly into trainNetwork, even without using arraydatastore. I typed 'edit trainNetwork' and read through the comments and this one explained a good amount of the steps:

% trainedNet = trainNetwork(X, Y, layers, options) trains and returns a

% network, trainedNet. The format of X depends on the input layer.

% - For an image input layer, X is a numeric array of images arranged

% so that the first three dimensions are the width, height and

% channels, and the last dimension indexes the individual images.

% - For a 3-D image input layer, X is a numeric array of 3-D images

% with the dimensions width, height, depth, channels, and the last

% dimension indexes the individual observations.