Skip to content

Latest commit

 

History

6 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Earth Surface Water Image Segmentation

We use satellite images from Sentinel 2 to identify which portions of the image correspond to water, using the dataset from Luo et al.'s 2021 paper.

For semantic segmentation, we use the U-Net++ architecture, with a Resnet34 backbone for our encoder. Although Sentinel-2 operates with 13 bands, Luo et al. in making their dataset have discarded the 7 lowest-resolution bands. Therefore, we have a 6-channel input images. Post-segmentation, each pixel will be classified as either water or land. Training was done from scratch, without use of pretrained layers.

Example Segmentations from the Trained Unet++ model

Segmentations

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages