As far as I have tried , you can reshape it as 2D and use it or you can use 1D as is

https://www.mathworks.com/matlabcentral/fileexchange/68882-cnn-classifier-using-1d-2d-and-3d-feature-vectors

We have tried this concept and the main issue that you could face is when your 1D feature vector exhibits some negative or complex values that don't adhere to images dynamic range. We did reshape HOG features, though, into a matrix then fed that to CNN.

Object recognition using shape growth pattern:

https://ieeexplore.ieee.org/abstract/document/8073567

Hi. If you are using MATLAB, you can simply use the reshape() function.

If you want your data to be in the range of a standard image, you can use mapminmax function to map your values between 0 and 1 or any other range.

You can use this command to make an uint8 image using a double image (an image with values between 0 and 1):

I= uint8(dob_im * 255);

I recommend you to examine the spectrogram too. The spectrogram is useful because it has the time and frequency information simultaneously and it's an image.

@Aparna Sathya Murthy, thanks for your information

@ Majid Farzaneh, thanks for the information

@majid farzaneh , thanks for the information