How can we give Hyperspectral image data which is in .mat file format and has more than 3 channels to a convolutional neural networks(CNN)?In website lot of tutorials are available which use colour image as input to CNNs but no resources are availlable for hyperspectral image containing reflectance values in .mat format.
If you use Matlab deep learning functions, you have to use an image input layer with nChannel greater than 3.
inputlayer = imageInputLayer([dim1 dim2 nChannel]);
https://it.mathworks.com/help/deeplearning/ref/nnet.cnn.layer.imageinputlayer.html?s_tid=doc_ta
Hi Ladi,
To add to Francesco's answer, I don't think have the data in .mat file matters much to the convolutional neural network. The first layer of the network will have to have the same number of channels as your input data (in your case hyperspectral images). In addition, the network will have to be re-trained on such images (with reflectance values) rather than pixel images (with RGB values).
Thanks for your answer.But can you tell me the process for pixel wise Classification.My image consists of multiple classes.
- Badges
- Science topic
- Similar topics
- Filing