The deep learning revolution has brought significant advances in a number of fields, primarily linked to image and speech recognition. The standardization of image classification tasks like the ImageNet Large Scale Visual Recognition Challenge has resulted in a reliable way to compare top performing architectures.

The module uses a deep learning image classifier with the Xception network using Keras on top of Tensorflow. The image classifier has been trained on the 10K most popular species from the iNaturalist dataset [inaturalist]. This dataset has around 4.4M observations with 7M images from 58K worldwide species.

― Biodiversity applications