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.
I have worked done this kind of task using python.Python can read the .mat data using scipy library.You can get the full information using the following link:
https://docs.scipy.org/doc/scipy/reference/tutorial/io.html
There are various algorithms for hyperspectral image classification like taking the pixel vector defining its neighborhood and assigning label to it using ground truth values.Please have a look to the following paper:
Article Deep&Dense Convolutional Neural Network for Hyperspectral Im...
If you have any other question related to this topic,feel free to contact me.
- Badges
- Science topic
- Similar topics
- Filing