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

Abstract

Abstract. Airborne and spaceborne platforms are the primary data sources for large-scale forest mapping, but visual interpretation for individual species determination is labor-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.

Monitor indicators of genetic diversity from space using Earth Observation data

“…globalforestwatch.org/) [35,13,30]. Public tools are increasingly available to classify tree species from EO data, especially for mono-dominant species in temperate forests, although high-spatial-resolution data (i.e., meter to sub-meter spatial data) are generally still required [1]. However, such classification results can be used to obtain a baseline estimate of N c , and detection of changes from this baseline can be achieved with EO products having lower spatial resolution e.g.…”

On the Effects of Different Types of Label Noise in Multi-Label Remote Sensing Image Classification

“…In the experiments, we have used three RS multi-label datasets: 1) DeepGlobe-ML, which is a multi-label dataset that we constructed from the DeepGlobe Land Cover Classification Challenge dataset [12], 2) BigEarthNet-S2 [18], and 3) TreeSatAI [62]. For each dataset, example images are shown in Figure 2.…”

“…3) TreeSatAI: The TreeSatAI dataset [62] is a multi-label tree species classification benchmark dataset that consists of 50,381 image-triplets of a high-resolution aerial image, a Sentinel-1 SAR and a Sentinel-2 multispectral image acquired over the German federal state of Lower Saxony. In our experiments, we only use the aerial images.…”

RDS - Recuperando Amostras Descartadas Com Rótulos Ruidosos: Técnicas Para Treinamento De Modelos De Deep Learning Com Amostras Ruidosas

“…Por se tratar de uma área em seus estágios iniciais de estudos, foi preciso criar dois conjuntos de dados multilabel com amostras ruidosas. Foi então gerada uma versão do dataset UcMerced [27] com amostras multilabel ruidosas, e uma versão do dataset TreeSatAI [28], também com amostras ruidosas. Esses datasets estão disponíveis em https://github.com/ICA-PUC, permitindo que trabalhos futuros realizem comparações nas mesmas condições e se torne um benchmark para novos avanços na área.…”

“…𝑀} sendo M ∈ 𝑁 + . Dado duas funções T(x) e T'(x) que aplica transformações distintas na entrada x, a amostra 𝑥 𝑖 transformada será dada por(28): logistic. A softmax com temperatura para amostra 𝑥 𝑖 e o modelo 𝑃(𝑦|𝑥: 𝜃) é dado por (30): 36 Sendo 𝛾 um hiperparâmetro utilizado para ponderar a relevância do 𝑅𝐷𝑆 𝐶𝐸_𝑃𝑒𝑠𝑜𝑠 na função de custo final do modelo.…”