VNIR Fusion
Spent time configuring the H7 project to handle loading an image
Due to the limited memory on the H7, I had to take one of the 500x700 (WxH) images and create successively smaller down-sampled versions of it
After testing 8 different images I found that the largest version of the image that can be loaded onto the H7 must be downsampled to 1/12 the original size which corresponds to an approximate size of 38x50 depending on the actual size of the input image
As of right now trouble lies in the conversion process for loading the image in that using the legacy code from the neural network project means it is possible being loaded in a grayscale format instead of RGB
I spend more than half a day trying to figure out how the H7 is reading in different pixel values than what MATLAB shows and I have not been able to figure it out yet. Still working on this
In the mean time, my partner, Bochi, is implementing the VNIR algorithm with the assumption that downsampled RGB imaged will be properly loaded into the H7
Comments