Skip to content

Estimate vegetation height from satellite images using deep learning.

License

Notifications You must be signed in to change notification settings

waldeland/deep_learning_forest_monitoring

Repository files navigation

Forest height estimation with deep learning

Code for the paper:
Forest mapping and monitoring of the African continent using Sentinel-2 data and deep learning,
Anders U. Waldeland, Øivind Due Trier, Arnt-Børre Salberg
International Journal of Applied Earth Observations and Geoinformation
Volume 111, July 2022, 102840
Paper is available here

Setup and prerequisites

Use a linux computer with CUDA installed and a nvidia-compatible GPU.

Install requirments

pip3 install -r requirements.txt

Usage

Predict vegetation height for a given Sentinel-2 product:

python3 predict_scene.py --product S2A_MSIL1C_20170126T073151_N0204_R049_T37MDQ_20170126T074339

Run clear cutting detection on a time-series of vegetation height estimates:

python3 forest_clear_cutting_detection.py

About

Estimate vegetation height from satellite images using deep learning.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages