Papers With Code 2 | ML Benchmarks, SotA Results & Code

WildlifeReID-10k is a wildlife re-identification dataset with more than 140k images of 10k individual animals. It is a collection of 37 existing wildlife re-identification datasets with additional processing steps. WildlifeReID-10k contains animals as diverse as marine turtles, primates, birds, African herbivores, marine mammals and domestic animals. We provide a Jupyter notebook with introduction to the dataset, a way to evaluate developed algorithms and a baseline performance. WildlifeReID-10k has two primary uses:

Design an algorithm to classify individual animals in images. This is the much more complicated task (with 10k fine-grained classes) and the intended use of the dataset.
Design an algorithm to classify species of animals. This is a simpler task (with 20 coarse-grained classes) requiring fewer resources. It is intended for researchers or interested public who want to develop their first methods on an interesting dataset.