iNaturalist Challenge at FGVC 2017

iNat Challenge 2017数据集包含5,089个物种，并结合了由67个图像组成的训练和验证集，这些图像已由inaturalist.org的多个用户收集并验证。该数据集包含许多视觉上相似的物种，这些物种在世界范围内的各种情况下均已捕获。

The iNat Challenge 2017 dataset contains 5,089 species, paired with training and validation sets consisting of 67 images. All these images were collected and verified by multiple users on inaturalist.org. This dataset includes numerous visually similar species captured in various contexts across the world.

搜集汇总

背景与挑战

背景概述

iNaturalist Challenge at FGVC 2017数据集包含5,089个物种，共计675,000张训练和验证图像，由inaturalist.org用户收集并验证。该数据集聚焦于全球范围内视觉相似物种的细粒度分类挑战。