A small application that gets and labels tarantulas by species. The labeled images can then be used to train an ML model to classify animals in new images.
- Get all currently accepted scientific names of tarantula species from the World Spider Catalog
- Get all formerly accepted scientific synonyms, as well
- Scrape Arachnoboards tags for images of each species + synonyms and label as the currently accepted scientific name
- Exclude unsuitable images (i.e. molts, enclosures, ventral shots, ...)
- Make separate directories that contain only those species with at least 10 and 100 images
- Is there a way to include gender?
- Scrape more sources
- Retrain model once there are more species with 100+ images