Model description

LightGBM classifier of tree/non-tree pixels from aerial imagery trained on Zurich's Orthofoto Sommer 2014/15 using detectree.

Intended uses & limitations

Segment tree/non-tree pixels from aerial imagery

Training Procedure

[More Information Needed]

Hyperparameters

Click to expand

Hyperparameter	Value
boosting_type	gbdt
class_weight
colsample_bytree	1.0
importance_type	split
learning_rate	0.1
max_depth	-1
min_child_samples	20
min_child_weight	0.001
min_split_gain	0.0
n_estimators	200
n_jobs
num_leaves	31
objective
random_state
reg_alpha	0.0
reg_lambda	0.0
subsample	1.0
subsample_for_bin	200000
subsample_freq	0

Model Plot

LGBMClassifier(n_estimators=200)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Evaluation Results

Metrics calculated on a validation set of 1% of the test tiles

Metric	Value
accuracy	0.87635
precision	0.785237
recall	0.756414
f1	0.770556

Dataset description

https://www.geolion.zh.ch/geodatensatz/2831

Preprocessing description

Images are resampled to 50 cm resolution. Train/test split based on image descriptors with 1% of tiles selected for training.

How to Get Started with the Model

[More Information Needed]

Model Card Authors

Martí Bosch

Model Card Contact

marti.bosch@epfl.ch

Citation

https://joss.theoj.org/papers/10.21105/joss.02172

Example predictions