Land Cover Classification Using Python, Remote Sensing Data, and Machine Learning
Hi geospatial enthusiast!
I would to share you something that I do to challenge myself. So far, I have mostly using Google Earth Engine for many geospatial project but I would also learn about using alternative without it.
Check out this repository: https://github.com/ramiqcom/lc-classi...
Please read the instruction and comment on the script
So this is a Python Notebook script to do land cover classification that I usually done on Google Earth Engine but instead on Python Environment. It is more complicated because instead of only 1 library you need to use multiple. Luckily most the library is compatible with each other. I am using multiple library such as geopandas, rasterio, scikit-learn, scikit-image, PIL, matplotlib, numpy, pandas, and more (in the dependencies).
On this modelling, I am using Landsat and Palsar satellite imagery as predictors, sample from a part of Indonesian Ministry of Life Environment and Forestry Land Cover data. I am using Random Forest model because it is very flexible and fast for this type of modelling.
You can see the perfomance of the model using assessment with confusion matrix and classification report, save the land cover map generated from the model, segment the land cover so that it is more generalize, and convert it as shapefile to your local drive.
This script is just one approach, you can modify it to your need. I hope this script is helpful on your project and research!
Best,
Ramadhan