iNat Challenge 2021
Overview
It is estimated that the natural world contains several million species of plants and animals. Without expert knowledge, many of these species are extremely difficult to classify due to their visual similarity. The goal of this competition is to push the state of the art in supervised image classification for real world data that features a large number of fine-grained categories.
The iNat Challenge 2021 dataset contains 10,000 species, with a training dataset of 2.7M images that have been collected and verified by multiple users from iNaturalist. There is also a more manageable "mini" dataset with 50 images per species for a total of 500K training images. The dataset features many visually similar species, captured in a wide variety of situations, from all over the world. Please refer to the iNaturalist 2021 Competition Github page for additional dataset details and download links.
Competition
Start Date - 8 March 2021
End Date - 28 May 2021
Kaggle URL - https://www.kaggle.com/c/inaturalist-2021
Github URL - https://github.com/visipedia/inat_comp/tree/master/2021
Winners
1st: BrownBlueGreendd
Fuhang Zong, Feng Zhu, Zhaoyan Liu, Yuan Tian, Xinglong Wu - ByteDance
2nd: amazing
Qishuai Diao - Beihang University (intern in ByteDance)
Organizers
Grant Van Horn (Cornell), Oisin Mac Aodha (University of Edinburgh)
Acknowledgements
AWS and iNaturalist