How to make the model work for images with 6 input channels?

[ShreyaPandita01]

trafficstars

My use case requires me to implement Unet for satellite imagery with additional inputs. So the 6 channels consist of first 3 RGB channels and then 3 channels consisting of the additional information I want to pass on to the Unet. How can I do that?

[jspaezp]

For what i can tell, most of the project is designed to work only on image inputs (as it is defined by a png image), so the input channels are constrained to 3 in the definition of the models.

You can take a look at other projects that implement segmentation, for instance https://github.com/jakeret/unet would be a good option