Skip to content

Darwin Core tables

The Darwin Core Event and Occurrence tables of the SmartWilds release are published as one Hugging Face dataset, ABC-Center/smartwilds_metadata, part of the SmartWilds collection. It holds the five components of the release, each a complete FAIR² package, plus the supplemental material:

smartwilds_metadata/
├── README.md                      # dataset card
├── SUPPLEMENTAL.md
├── smartwilds-camera-traps/       # 19,785 Events, 72,775 Occurrences, 235,778-image inventory
├── smartwilds-bioacoustics/       # 2,382 Events, 9,960 Occurrences, 2,371 recordings
├── smartwilds-drones/             # 55 Events, 318,215 Occurrences, frame-indexed detections
├── smartwilds-citizen-science/    # 14 Events, 622 Occurrences
└── smartwilds/                    # study level: merged tables, text records, arrival traces
    └── <component>/
        ├── README.md              # component dataset card
        ├── data/                  # events.csv, occurrences.csv, per-file tables
        ├── metadata/              # DATA_DICTIONARY.md, deployments, configs, provenance/
        ├── media_manifest.csv
        └── CHECKSUMS.sha256

Release build 2026-10-22; camera and merged tables updated 2026-11-08. All tables share one Event hierarchy rooted at TheWilds_2025.

Getting the tables

pip install huggingface_hub
huggingface-cli download --repo-type dataset ABC-Center/smartwilds_metadata --local-dir smartwilds_metadata
python examples/quickstart.py smartwilds_metadata/smartwilds-camera-traps smartwilds_metadata/smartwilds-bioacoustics

Or load one table directly:

import pandas as pd
oc = pd.read_csv("hf://datasets/ABC-Center/smartwilds_metadata/smartwilds-camera-traps/data/occurrences.csv",
                 dtype=str, keep_default_na=False)

Column definitions

Each component's metadata/DATA_DICTIONARY.md defines its columns. The Darwin Core terms follow https://dwc.tdwg.org/terms/; SmartWilds-specific columns (labelProvenance, geometryProvenance, triageRound, the timestamp correction fields of images.csv, the privacy screening fields of the media manifests) are defined in the supplemental material.