LILA BC — Community Mirror (unofficial)

HF-native, viewer-ready object-detection versions of camera-trap datasets from the Labeled Information Library of Alexandria: Biology and Conservation (LILA BC).

⚠️ This is an unofficial community mirror — not affiliated with or endorsed by LILA BC. All datasets retain their original CDLA-Permissive-1.0 license and citations. LILA maintainers are warmly welcome to adopt, rename, or take ownership of this org.

Why this exists

The datasets are materialised into parquet so that:

• 🖼️ the Hugging Face dataset viewer renders the bounding boxes, and
• 🚀 each set trains in one command with the vision-trainer skill.

This complements the existing LILA presence on the Hub rather than competing with it:

Datasets

• ena24-detection — Eastern North America, 23 species (species-level boxes)
• channel-islands-camera-traps — Channel Islands, CA (The Nature Conservancy); fox/bird/rodent/skunk/other
• missouri-camera-traps — 20 species (boxed subset)

Built with the open build_lila_detection.py converter (resolves any LILA dataset from the official metadata table). Empty frames and privacy-stripped human labels are excluded so the sets are clean for detection.

Maintained by @davanstrien. Questions or a request to take this over? Open a discussion on any dataset here, or reach LILA at info@lila.science.