search

SatelliteShorelines / CoastSeg

An interactive toolbox for downloading satellite imagery, applying image segmentation models, mapping shoreline positions and more. The mapping extension for CoastSat and Zoo.
https://satelliteshorelines.github.io/CoastSeg/
GNU General Public License v3.0
43 stars 6 forks source link

Zoo now has new Zenodo model and better (and faster) programmatic access to those models. Suggestions for how to incorporate in CoastSeg #92

Closed dbuscombe-usgs closed 1 year ago

dbuscombe-usgs commented 1 year ago

Updated code See https://github.com/Doodleverse/segmentation_zoo/issues/15 and the rewritten select_model_and_batch_process_folder.py script

CoastSeg could be modified along these lines

Updated models The wiki has been updated with details of all 5 of the new available models on Zenodo https://github.com/Doodleverse/segmentation_zoo/wiki/4_Models#landsatsentinel-2-models

Note that these are only for 4-class (water, whitewater, sediment, other)

I am now working on the simpler 2-class (water, nowater) models

According to the changes I suggested in https://github.com/SatelliteShorelines/CoastSeg/issues/89, the forthcoming 2-class models should be made available in SDS_coastsat_classifier.ipynb only, and both the 4-class and the 2-class models should be made available in SDS_unet_classifier.ipynb

2320sharon commented 1 year ago

I've implemented a function to convert RGB imagery to MNDWI and NDWI imagery in the commit https://github.com/SatelliteShorelines/CoastSeg/commit/8c6dc68bfc9e21c0c203adc3d0ef5602fa4772ab When the user clicks run model the RGB directory along with the NIR and SWIR directories are automatically found and used to create a new directory 'MNDWI '/'NDWI' which contains the newly created imagery.