Going from computer vision to remote sensing, the first thing that strikes you is there are more than 3 bands/channels. The pixel values are not between 0 and 255! Madness!
Some of these wavelengths are not visible to the human eye. Yet, we can still use them to train ML models.
I'm using all 12 bands to train a U-Net used to detect the coastline. Which bands do you think would be most important to the prediction?
