# Detectree2 Crown Data, CanopyRS Preprocessed Tiles: Manually Delineated Tropical Tree Crowns on Aerial RGB of Danum and Sepilok, Malaysian Borneo

## Facts

- Provider: Ball, Jackson, Hickman and Koay; preprocessed by the CanopyRS authors
- License: CC-BY-4.0 (https://zenodo.org/records/8136161)
- Size: 6 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/CanopyRS/Detectree2
- Access: public, readable without an account
- DOI: https://doi.org/10.5281/zenodo.8136161
- Category: Environmental Science
- Source: https://huggingface.co/datasets/CanopyRS/Detectree2
- Page: https://scigantic.com/archive/dc628baa-2340-5adc-b4f2-1e9c28566360
- Keywords: Environmental Science, image segmentation, Parquet, Hugging Face

## About

A tiled version of the crown data behind detectree2 (Ball, Jackson, Hickman and Koay), prepared by the CanopyRS authors for the SelvaBox tree crown detection study. The original deposit pairs airborne (aeroplane-mounted) RGB orthomosaics of tropical rainforest at Danum Valley and Sepilok (East and West) in Sabah, Malaysian Borneo with manually delineated tree crown polygons, used to train the Mask R-CNN based detectree2 crown delineation model. Here the imagery is cut into tiles with COCO-format crown annotations, split into 331 validation and 311 test tiles, stored as Parquet shards under data/ (about 1.5 GB) and loadable with datasets.load_dataset. Useful for evaluating tree crown detection and instance segmentation in Bornean tropical forest and for cross-continent transfer of canopy models. The original GeoPackage crowns, GeoTIFF orthomosaics and predictions are on Zenodo (DOI 10.5281/zenodo.8136161). Original data licensed CC BY 4.0; cite Ball et al. (2023), doi:10.1002/rse2.332, and the SelvaBox paper.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("CanopyRS/Detectree2")
print(ds)
```

## Data structure

File types in sampled files: Parquet

Example file: `data/test-00000-of-00002.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/dc628baa-2340-5adc-b4f2-1e9c28566360
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/dc628baa-2340-5adc-b4f2-1e9c28566360
- Over MCP: `get_schema_card` with id `dc628baa-2340-5adc-b4f2-1e9c28566360` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Image segmentation datasets](https://scigantic.com/topics/image-segmentation)
- [Environmental Science datasets](https://scigantic.com/topics/environmental-science)
