Preprints
https://doi.org/10.5194/essd-2022-312
https://doi.org/10.5194/essd-2022-312
 
26 Sep 2022
26 Sep 2022
Status: this preprint is currently under review for the journal ESSD.

TreeSatAI Benchmark Archive: A multi-sensor, multi-label dataset for tree species classification in remote sensing

Steve Ahlswede1,, Christian Schulz2,, Christiano Gava3, Patrick Helber4, Benjamin Bischke4, Michael Förster2, Florencia Arias2, Jörn Hees3, Begüm Demir1, and Birgit Kleinschmit2 Steve Ahlswede et al.
  • 1Technische Universität Berlin, Germany, Remote Sensing Image Analysis Group, Einsteinufer 17, 10587 Berlin, Germany
  • 2Technische Universität Berlin, Geoinformation in Environmental Planning Lab, Straße des 17. Juni 145, 10623 Berlin, Germany
  • 3Deutsches Forschungszentrum für Künstliche Intelligenz GmbH (DFKI), Smart Data and Knowledge Services, Trippstadter Str. 122, 67663 Kaiserslautern, Germany
  • 4Vision Impulse GmbH, Trippstadter Str. 122, 67663 Kaiserslautern, Germany
  • These authors contributed equally to this work.

Abstract. Airborne and spaceborne platforms are the primary data sources for large-scale forest mapping, but visual interpretation for individual species determination is labour-intensive. Hence, various studies focusing on forests have investigated the benefits of multiple sensors for automated tree species classification. However, transferable deep learning approaches for large-scale applications are still lacking. This gap motivated us to create a novel dataset for tree species classification in Central Europe based on multi-sensor data from aerial, Sentinel-1 and Sentinel-2 imagery.

In this paper, we introduce the TreeSatAI Benchmark Archive, which contains labels of 20 European tree species (i.e., 15 tree genera) derived from forest administration data of the federal state of Lower Saxony, Germany. We propose models and guidelines for the application of the latest machine learning techniques for the task of tree species classification with multi-label data. Finally, we provide various benchmark experiments showcasing the information which can be derived from the different sensors including artificial neural networks and tree-based machine learning methods.

We found that residual neural networks (ResNet) perform sufficiently well with weighted precision scores up to 79 % only by using the RGB bands of aerial imagery. This result indicates that the spatial content present within the 0.2 m resolution data is very informative for tree species classification. With the incorporation of Sentinel-1 and Sentinel-2 imagery, performance improved marginally. However, the sole use of Sentinel-2 still allows for weighted precision scores of up to 74 % using either multi-layer perceptron (MLP) or Light Gradient Boosting Machine (LightGBM) models. Since the dataset is derived from real-world reference data, it contains high class imbalances. We found that this dataset attribute negatively affects the models' performances for many of the underrepresented classes (i.e., scarce tree species). However, the class-wise precision of the best performing late fusion model still reached values ranging from 54 % (Acer) to 88 % (Pinus). Based on our results, we conclude that deep learning techniques using aerial imagery could considerably support forestry administration in the provision of large-scale tree species maps at a very high resolution to plan for challenges driven by global environmental change.

The original dataset used in this paper is shared via Zenodo (https://doi.org/10.5281/zenodo.6598390) [Schulz et al., 2022]. For citation of the dataset, we refer to this article.

Steve Ahlswede et al.

Status: open (until 07 Dec 2022)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on essd-2022-312', Anonymous Referee #1, 03 Oct 2022 reply
    • AC1: 'Reply on RC1', Christian Schulz, 23 Nov 2022 reply
  • RC2: 'Comment on essd-2022-312', Anonymous Referee #2, 27 Nov 2022 reply

Steve Ahlswede et al.

Data sets

TreeSatAI Benchmark Archive for Deep Learning in Forest Applications Schulz, Christian; Ahlswede, Steve; Gava, Christiano; Helber, Patrick; Bischke, Benjamin; Arias, Florencia; Förster, Michael; Hees, Jörn; Demir, Begüm; Kleinschmit, Birgit https://doi.org/10.5281/zenodo.6598390

Model code and software

treesatai_benchmark Gava, Christiano; Hees, Jörn https://github.com/DFKI/treesatai_benchmark

TreeSat_Benchmark Ahlswede, Steve; Demir, Begüm https://git.tu-berlin.de/rsim/treesat_benchmark

Steve Ahlswede et al.

Viewed

Total article views: 548 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
391 141 16 548 9 7
  • HTML: 391
  • PDF: 141
  • XML: 16
  • Total: 548
  • BibTeX: 9
  • EndNote: 7
Views and downloads (calculated since 26 Sep 2022)
Cumulative views and downloads (calculated since 26 Sep 2022)

Viewed (geographical distribution)

Total article views: 528 (including HTML, PDF, and XML) Thereof 528 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 30 Nov 2022
Download
Short summary
Imagery from air and space is the primary source for large-scale forest mapping. Our study introduces a new dataset with over 50,000 image patches prepared for deep learning tasks. We show how the information for 20 European tree species can be extracted from different remote sensing sensors. Our algorithms can detect single species with precision scores up to 88 %. With a pixel size of 20x20 cm, forestry administration can now derive large-scale tree species maps at a very high resolution.