Measured performance improvement of MSM (2023) vs Observation model (2021) using the same test data. Accuracy is measured as the percentage of observations in which the first prediction is correct. Average recall is measured as the average recognition rate per taxon. Higher values indicate better recognition of rarer taxa.

  
  Part of: Hogeweg L, Yan N, Brunink D, Ezzaki-Chokri K, Gerritsen W, Pucci R, Ghani B, Stowell D, Kalkman VJ (2024) AI Species Identification Using Image and Sound Recognition for Citizen Science, Collection Management and Biomonitoring: From Training Pipeline to Large-Scale Models. Biodiversity Information Science and Standards 8: e136839. https://doi.org/10.3897/biss.8.136839