SuAVE for Biodiversity and Ecology

We have published several camera trap and biodiversity image collections in SuAVE. Below you will find examples from iDigBio and the Florida Museum of Natural History, iNaturalist, and the TEAM project. A range of questions can be addressed by exploring these collections, such as species abundance, activity of different species by time of day and temperature.

Camera traps. This collection of camera trap images of wildlife in tiger habitats from the Thailand DNP was used during the 2014 workshop on Tiger Habitat Conservation in Bangkok. Several other camera trap collections have been published earlier, such as the Carnivora: Terrestrial Vertebrate Collection, from Central Suriname Nature Reserve, 2008, maintained by the TEAM project.
iNaturalist observations. This is the AfriBat collection assembled by iNaturalist members. You can look at the distribution of bat species in Africa (eg search for Giant bat) and explore phenotypes.
Machine learning for taxonomic identification.Butterfly images from the Florida Museum of Natural History. This image collection is used to develop and test a Convolutional Neural Network (CNN) model for taxonomic identification of butterfly species. A Jupyter notebook implementing CNN is called directly from SuAVE; it is trained to predict "Common Name" values, with 75% of images used for training and 25% for testing. The predicted values of "Common name", along with several additional model variables, are then saved as a new SuAVE survey. In the survey on the left, visualize the confusion matrix by cross-tabulating the original and predicted common names.
Galleries of species.Fish and Shark Profiles from the Florida Museum of Natural History. There is also a larger fish gallery from the same museum.

Additional collections to explore