Landcover classification of drone UAV imagery (RGB and Multispectral) using Random forest with unsupervised segmentation (superpixel) postprocessing.
In this repository I only store the code (Python). The data is taken from Hilden collection, which is accessible by contacting the Hilden Network.
The code is organized into 4 Jupyter notebooks and one Python script with functions:
- 0._ - preprocessing of the images. Filling the gaps and resizing the images if needed.
- 1._ - showing an example of landcover classification on one image
- 2._ - classifying all the prepared in script 0. multispectral imagery
- 3._ - classifying all the prepared in script 0. RGB imagery
- myfunctions - all the custom functions I used
If you want to quickly glance at how the analysis is done, just view the 1._script. It contains images and descriptions of each step.
Please feel free to contact me if you have any questions!