Deep Learning for Satellite Imagery

I am working on an image classification problem using satellite imagery.
I am using the Sentinel TOA reflectance images which has up to 13 bands, I want to use 5 of them for my model. Since I am using conv nets I was wondering if there is a way of using transfer learning.

Most of the pretrained models on satellite data of sentinel are available for 13 band images, I am not able to figure how I can use them on 5 band images. Is there a way to configure the weights?


1 Like