Land Use and Land Cover Classification using a ResNet Deep Learning Architecture

Hello, I have developed a Jupyter notebook about Land Use and Land Cover classification using a pretrained ResNet50 architecture finetuned with the EuroSAT dataset from the Copernicus Sentinel-2 products. I didn’t need anything more than the fastai library to write the code. If you are interested in using deep learning for satellite images, the notebook is available here:

Enjoy !


Excellent write up, thanks for sharing

Thanks Robin.

1 Like

few days ago I have finished a small project about using a ResNet architecture finetuned with Sentinel-2 patch images for Land Use and Land Cover classification. It’s not about SNAP but it can be interesting for those who are working on ConvNets with satellite images. Here is the link to an introductory post and from there a link to the Jupyter notebook.