Urban tree species benchmark dataset for time series classificationEasyData - Data Terra
Classification of urban tree species is essential for understanding their ecological functions, managing urban forests (public and private), and informing nature-based solutions for climate resilience. We present a benchmark dataset for urban tree species classification based on multi-source optical...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-08-01
|
| Series: | Data in Brief |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2352340925005049 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Classification of urban tree species is essential for understanding their ecological functions, managing urban forests (public and private), and informing nature-based solutions for climate resilience. We present a benchmark dataset for urban tree species classification based on multi-source optical satellite image time series (SITS). The dataset provides, on the city of Strasbourg (France), surface reflectance values extracted from coregistered Sentinel-2 and PlanetScope imagery on public trees. Species labels and geolocations are derived from the city inventory Patrimoine arboré 2022. A total of 45,084 trees representing the 20 most common species are included. The dataset is formatted for time series classification, with surface reflectance values and consistent spatial sampling. It supports direct integration into deep learning frameworks and includes three InceptionTime-based models trained on Sentinel-2, PlanetScope and both sources through a fusion architecture (Dual-InceptionTime). Model outputs—predicted species, confidence scores, and correctness flags—are provided, along with an interactive t-SNE visualization of the latent feature space for interpretability and error analysis. This dataset offers a reproducible framework for evaluating species classification models, fusion strategies, and explainability techniques, and contributes to advancing urban vegetation monitoring using satellite image time series and deep learning. |
|---|---|
| ISSN: | 2352-3409 |