Researchers Build Massive Dataset and AI to Better Classify Organisms
• Researchers developed a dataset of over 10 million biological images for machine learning, significantly expanding AI capabilities.
• They also created BioCLIP, an AI model that learns from images and text to classify organisms.
• BioCLIP performed 17-20% better than existing models at classifying images in the tree of life.
• The model succeeds by noticing subtle visual differences between similar species.
• Future versions could incorporate more scientific data to further improve AI taxonomic capabilities.