/canopy

Automatic tree species classification from remote sensing data

Primary LanguagePythonMIT LicenseMIT

canopy

Automatic tree species classification from remote sensing data

Code from our paper:

Fricker, G. A., Ventura, J. D., Wolf, J. A., North, M. P., Davis, F. W., & Franklin, J. (2019). A Convolutional Neural Network Classifier Identifies Tree Species in Mixed-Conifer Forest from Hyperspectral Imagery. Remote Sensing, 11(19), 2326.

To create a conda environment:

conda create -n canopy tensorflow=1.10.0 pip ;
conda activate canopy ;
pip3 install -r docker/requirements.txt ;

To run the example experiment:

mkdir hyperspectral ; 
python -m experiment.download ;
python -m experiment.preprocess --out hyperspectral;
python -m experiment.train --out hyperspectral;
python -m experiment.test --out hyperspectral;
python -m experiment.analyze --out hyperspectral;