Lidar-based Norwegian tree species detection using deep learning

Martijn Vermeer, Jacob Alexander Hay, David Völgyes, Zsofia Koma, Johannes Breidenbach, Daniele Fantin
Proceedings of the 5th Northern Lights Deep Learning Conference ({NLDL}), PMLR 233:228-234, 2024.

Abstract

Background: The mapping of tree species within Norwegian forests is a time-consuming process, involving forest associations relying on manual labeling by experts. The process can involve aerial imagery, personal familiarity, on-scene references, and remote sensing data. The state-of-the-art methods usually use high-resolution aerial imagery with semantic segmentation methods. Methods: We present a deep learning based tree species classification model utilizing only lidar (Light Detection And Ranging) data. The lidar images are segmented into four classes (Norway Spruce, Scots Pine, Birch, background) with a U-Net based network. The model is trained with focal loss over partial weak labels. A major benefit of the approach is that both the lidar imagery and the base map for the labels have free and open access. Results: Our tree species classification model achieves a macro-averaged $\mathrm{F}_1$ score of 0.70 on an independent validation with National Forest Inventory (NFI) in-situ sample plots. That is close to, but below the performance of aerial, or aerial and lidar combined models.

Cite this Paper

BibTeX
@InProceedings{pmlr-v233-vermeer24a, title = {Lidar-based Norwegian tree species detection using deep learning}, author = {Vermeer, Martijn and Hay, Jacob Alexander and V{\"o}lgyes, David and Koma, Zsofia and Breidenbach, Johannes and Fantin, Daniele}, booktitle = {Proceedings of the 5th Northern Lights Deep Learning Conference ({NLDL})}, pages = {228--234}, year = {2024}, editor = {Lutchyn, Tetiana and Ramírez Rivera, Adín and Ricaud, Benjamin}, volume = {233}, series = {Proceedings of Machine Learning Research}, month = {09--11 Jan}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v233/vermeer24a/vermeer24a.pdf}, url = {https://proceedings.mlr.press/v233/vermeer24a.html}, abstract = {Background: The mapping of tree species within Norwegian forests is a time-consuming process, involving forest associations relying on manual labeling by experts. The process can involve aerial imagery, personal familiarity, on-scene references, and remote sensing data. The state-of-the-art methods usually use high-resolution aerial imagery with semantic segmentation methods. Methods: We present a deep learning based tree species classification model utilizing only lidar (Light Detection And Ranging) data. The lidar images are segmented into four classes (Norway Spruce, Scots Pine, Birch, background) with a U-Net based network. The model is trained with focal loss over partial weak labels. A major benefit of the approach is that both the lidar imagery and the base map for the labels have free and open access. Results: Our tree species classification model achieves a macro-averaged $\mathrm{F}_1$ score of 0.70 on an independent validation with National Forest Inventory (NFI) in-situ sample plots. That is close to, but below the performance of aerial, or aerial and lidar combined models.} }
Endnote
%0 Conference Paper %T Lidar-based Norwegian tree species detection using deep learning %A Martijn Vermeer %A Jacob Alexander Hay %A David Völgyes %A Zsofia Koma %A Johannes Breidenbach %A Daniele Fantin %B Proceedings of the 5th Northern Lights Deep Learning Conference ({NLDL}) %C Proceedings of Machine Learning Research %D 2024 %E Tetiana Lutchyn %E Adín Ramírez Rivera %E Benjamin Ricaud %F pmlr-v233-vermeer24a %I PMLR %P 228--234 %U https://proceedings.mlr.press/v233/vermeer24a.html %V 233 %X Background: The mapping of tree species within Norwegian forests is a time-consuming process, involving forest associations relying on manual labeling by experts. The process can involve aerial imagery, personal familiarity, on-scene references, and remote sensing data. The state-of-the-art methods usually use high-resolution aerial imagery with semantic segmentation methods. Methods: We present a deep learning based tree species classification model utilizing only lidar (Light Detection And Ranging) data. The lidar images are segmented into four classes (Norway Spruce, Scots Pine, Birch, background) with a U-Net based network. The model is trained with focal loss over partial weak labels. A major benefit of the approach is that both the lidar imagery and the base map for the labels have free and open access. Results: Our tree species classification model achieves a macro-averaged $\mathrm{F}_1$ score of 0.70 on an independent validation with National Forest Inventory (NFI) in-situ sample plots. That is close to, but below the performance of aerial, or aerial and lidar combined models.
APA
Vermeer, M., Hay, J.A., Völgyes, D., Koma, Z., Breidenbach, J. & Fantin, D.. (2024). Lidar-based Norwegian tree species detection using deep learning. Proceedings of the 5th Northern Lights Deep Learning Conference ({NLDL}), in Proceedings of Machine Learning Research 233:228-234 Available from https://proceedings.mlr.press/v233/vermeer24a.html.

Related Material