HirakoSan/bio-lens

🌿 iNaturalist Silver Dataset (Research-Grade, Deduplicated)

Description

This dataset contains research-grade observations from iNaturalist, processed through a bronze-layer to silver pipeline that includes:

• Research-grade only observations (community-verified)
• One photo per observation (deduplicated by observation_uuid)
• AVIF encoded images stored as binary in Parquet
• ~ 5.6 TB total size across > 40k shards

Data Composition

Metric	Count
Total photos	~142.6M
Research-grade observations	~142.8M
Unique families with photos	~6,890
Unique taxa	5~05K
Shards	~41,315 Parquet files
Shard size	~500MB each

Top Families by Photo Count

Family	Photos
Asteraceae (daisies/sunflowers)	11.7M
Fabaceae (legumes)	6.4M
Nymphalidae (brush-footed butterflies)	5.6M
Anatidae (ducks/geese)	4.8M
Rosaceae (roses)	3.6M
Apidae (bees)	3.4M
Lamiaceae (mint family)	3.0M
Poaceae (grasses)	2.9M
Accipitridae (hawks/eagles)	2.9M
Orchidaceae (orchids)	2.7M

File Structure

data/train/
├── 0000.parquet
├── 0001.parquet
├── ...
└── 41315.parquet

Schema

Column	Type	Description
photo_id	int64	Unique photo identifier
observation_uuid	string	UUID (deduplicated, one per observation)
taxon_id	int64	iNaturalist taxon ID
taxon_name	string	Scientific name
ancestry	string	Taxonomic lineage path
rank	string	species, genus, family, etc.
image	binary	AVIF encoded image
license	string	Photo license
obs_quality_grade	string	Always "research"
latitude	double	Observation latitude
longitude	double	Observation longitude

Image Format

• Format: AVIF (AV1 Image File Format)
• Quality: 75
• Dimensions: Original (preserved)
• Average size: ~15-30 KB
• Metadata: Stripped (EXIF/XMP removed)

Licenses

Photos include licenses as selected by observers: CC0, CC-BY, CC-BY-NC, CC-BY-SA, CC-BY-NC-SA, CC-BY-ND, CC-BY-NC-ND.

⚠️ Non-commercial research use only

Usage

import polars as pl

# Load all data
df = pl.read_parquet("data/train/*.parquet")

# Filter by family
beetles = df.filter(pl.col("family") == "Carabidae")

# Access image
row = df.first()
image_bytes = row['image']

Processing Pipeline

This dataset was created by:

1. Joining photos, observations, and taxonomy
2. Filtering to research-grade observations
3. Deduplicating to one photo per observation
4. Converting images to AVIF
5. Sharding into ~500MB Parquet files

Citation

@misc{bio-lens,
  author = {iNaturalist Contributors, HirakoSan},
  title = {bio-lens},
  year = {2026}
}