TreeOfLife-200M provides nearly 214 million images representing 952,257 taxa across the tree of life. It is the largest public ML-ready dataset for computer vision in biology, compiled from four biodiversity data providers including GBIF and Encyclopedia of Life.
Use Cases
- Train a vision model on 214 million images to classify species across 952,257 taxa.
- Analyze image diversity and representation across the tree of life using the taxa labels.
- Develop models for biodiversity monitoring by leveraging the combined data from GBIF, EOL, BIOSCAN-5M, and FathomNet sources.
Strengths
- Contains nearly 214 million images, offering massive scale for model training.
- Covers a diverse range of 952,257 unique taxa across the biological tree of life.
- Combines data from four established biodiversity providers: GBIF, EOL, BIOSCAN-5M, and FathomNet.
Limitations
- Dataset size and complexity may require significant computational resources for processing.
- Image quality and annotation consistency may vary across the four source providers.
Provenance
- Source
- Global Biodiversity Information Facility (GBIF), Encyclopedia of Life (EOL), BIOSCAN-5M, FathomNet
- Collection Method
- Combined images and metadata from four core biodiversity data providers.
- Time Range
- null
- Freshness
- null
- Geography
- null