The Semi-Supervised iNaturalist Challenge at the FGVC8 Workshop
This dataset addresses the problem of semi-supervised learning with complex real-world data for computer vision researchers, but it is incremental as it builds on a previous challenge with expanded scale and taxonomic features.
The paper introduces the Semi-iNat dataset for semi-supervised classification, featuring a long-tailed distribution, fine-grained categories, and domain shifts, with 810 in-class and 1629 out-of-class species totaling 330k images, and provides baseline results for a challenge at CVPR 2021.
Semi-iNat is a challenging dataset for semi-supervised classification with a long-tailed distribution of classes, fine-grained categories, and domain shifts between labeled and unlabeled data. This dataset is behind the second iteration of the semi-supervised recognition challenge to be held at the FGVC8 workshop at CVPR 2021. Different from the previous one, this dataset (i) includes images of species from different kingdoms in the natural taxonomy, (ii) is at a larger scale -- with 810 in-class and 1629 out-of-class species for a total of 330k images, and (iii) does not provide in/out-of-class labels, but provides coarse taxonomic labels (kingdom and phylum) for the unlabeled images. This document describes baseline results and the details of the dataset which is available here: \url{https://github.com/cvl-umass/semi-inat-2021}.