metadata
metrics:
- accuracy
- f1
pipeline_tag: image-segmentation
tags:
- geospatial
- forest
- species
- productivity
- age
Models for "Characterizing long-term boreal forest ecosystems dynamics in Khanty-Mansi Autonomous Okrug (Western Siberia) using time series remote sensing data and deep learning methods"
Paper
Paper is published at Forest Ecology and Management and is available at (link here)
Other data
Jupyter notebooks, statistics and model logs are published on GitHub, maps are published on Zenodo
Models description
This repository contains SegFormer model checkpoints of 5 models that model dominant tree species, forest productivity class, forest type, age and growing stock in boreal forests of West Siberia.
The checkpoints are created with Remote Sensing Processor library, but can be loaded with Lightning too.