creating a cell array for classifying images using the trained stacked autoencoders

Question

Sivaramakrishnan Rajaraman 2017-3-19

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/330676-creating-a-cell-array-for-classifying-images-using-the-trained-stacked-autoencoders

回答： Radians 2019-12-7

Hi, I am having a folder of 10000 images (64*64 double, .bmp format), say at C:\images. How can I create a cell array of 1*10000 cell (each cell containing a 64*64 double image) and name the cell array as "digitaldata"?

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

Renee Coetsee 2017-3-23

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/330676-creating-a-cell-array-for-classifying-images-using-the-trained-stacked-autoencoders#answer_260082

Since you have so many images, I recommend checking out "ImageDatastore". You input a path to a folder of images and it will not put all the images into memory. If you store each image in a cell array, it will use a lot of memory. Here is a link to the documentation page for "ImageDatastore":

https://www.mathworks.com/help/matlab/ref/imagedatastore-object.html

To load the images into a cell array, you can use a for loop and the "imread" function.

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。

Answer 2

Jason Vann 2018-2-21

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/330676-creating-a-cell-array-for-classifying-images-using-the-trained-stacked-autoencoders#answer_306315

ImageDatastore cannot be used as an input to an Autoencoder (as of R2017b), even though it can be used with CNNs. Is there workaround for this issue? I guess one could theoretically put an autoencoder in a loop (saving all of the weights and biases) then continually load a new image matrices.