Source code for paper related to forest species mapping supported with machine learning using spectral, terrain and texture features
We observed the influence of various feature types, including values of initial bands, vegetation indices, terrain features and texture features on the performance of popular machine learning algorithms, namely Random Forest, Support Vector Machine, eXtreme Gradient Boosting and kNN
Map demonstrate classified forest map by Ridge Regression
Clone this repository
Install python packages
- matplotlib
- numpy
- seaborn
- sklearn
- pandas
- ipython
- jupyter
requirement - Docker ๐ณ
- Clone repository to your local machine
git clone https://github.com/mishagrol/MLForestMapping.git
- Go to folder
cd MLForestMapping
- Run Jupyter in Docker
bash run_in_docker.sh
- Open Jupyter in browser at
localhost:8890
, token isSecretToken
Availdable by request - mikhail.gasanov[@]skoltech.ru
2019 year data Y.Disk
2020 year data Y.Disk
Freely available SRTM
Texture features generated by EO-learn
package - https://github.com/sentinel-hub/eo-learn
Model_Train.ipynb
- jupyter to conduct ML model training with satellite data, spectral, texture and terrain features
Inference.ipynb
- jupyter to conduct inference of best models to recieve pixel-wise forest and UQ maps
SHAP.ipynb
- jupyter to conduct SHAP-analysis of pretrained ML models
BarPlots, Maps and other visualization created with python.
To reproduce plots install python package
- SciencePlots
Distributed under the MIT license. See LICENSE
for more information.