Machine learning training data: over 500,000 images of butterflies and moths (Lepidoptera) with species labels

The dataset contains 541,677 images of 185 butterfly and moth species that were recorded by citizen scientists in Austria with the application "Schmetterlinge Österreichs" of the foundation Blühendes Österreich (https://www.schmetterlingsapp.at/) between 2016 and 2023. Each image is identified on species level (with the exception of four species pairs). Correct identification was confirmed by an experienced entomologist. The images are organized in indiviudal .zip files for each species. The file pytorch_model.bin contains the weights of a MaxVit-t model (Tu et al., 2022) that was trained on species identification with the dataset. The file scripts_model_training.zip contains the scripts which was used for model training and testing on the EuroHPC supercomputer LUMI hosted by CSC (Finland) and the LUMI consortium. The file images_per_species.csv contains a list with the species in the dataset and the number of images of each species.More information on the dataset and model training are available in:<br>Barkmann, F., Lindner, A., Würflinger, R., Höttinger, H., Rüdisser, J., 2025. Machine learning training data: over 500,000 images of butterflies and moths (Lepidoptera) with species labels. <i>Sci Data 12</i>(1), 1369. https://doi.org/10.1038/s41597-025-05708-z