CANNOT make an image prediction after trainning a CNN network

Question

Jie 2022-11-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1847828-cannot-make-an-image-prediction-after-trainning-a-cnn-network

回答： Sanjana 2023-6-7

After I trained the CNN following the Matlab document, I get a net variable which represents the network. However when I tried to use it to predict new pictures, I read the image which size is 128 by 128 and used net(imgTEST), however it firstly hinted that Array indices must be positive integers or logical values. So I binarized the image and net(imgTEST) again, and then it reported that The logical indices contain a true value outside of the array bounds.

The sizes of pictures I used to train the network are also 128 by 128, so I don't know where goes wrong. Could somebody help? Thank you!

My net's property is given below.

net =

SeriesNetwork with properties:

Layers: [14×1 nnet.cnn.layer.Layer]

InputNames: {'imageinput'}

OutputNames: {'classoutput'}

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

Sanjana 2023-6-7

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1847828-cannot-make-an-image-prediction-after-trainning-a-cnn-network#answer_1251629

Hi Jie,

I understand that you are encountering some difficulties with performing predictions on a Test Image with the trained CNN Network.

To perform predictions with a trained CNN Network, which is usually a “dlnetwork” or “DAGNetwork”, you can use “predict” function, with inputs as the network obj and input resized to the size of the network input.

Please refer to the following links, for further information,

https://www.mathworks.com/help/deeplearning/ug/make-predictions-using-dlnetwork-object.html

https://www.mathworks.com/help/deeplearning/ref/dlnetwork.predict.html

Hope this helps!