Ecuador moth dataset, Costa rica dataset

本研究介绍了两个用于生物多样性分析的细粒度视觉分类数据集：Ecuador moth dataset和Costa rica dataset。Ecuador moth dataset包含675种在南厄瓜多尔山区热带雨林中量化收集的尺蛾科昆虫，数据集挑战性极高，包含许多科学上未知的相似物种。Costa rica dataset则涵盖了哥斯达黎加西北部广泛收集的多种蛾和蝴蝶，数据集已公开，并计划发布转换后的元数据以方便计算机视觉社区使用。两个数据集均用于解决生物多样性研究中的物种识别和分类问题，通过自动化视觉系统辅助专家进行物种区分和注释，以及对物种差异进行系统性和定量分析。

This study introduces two fine-grained visual classification datasets for biodiversity analysis: the Ecuador Moth Dataset and the Costa Rica Dataset. The Ecuador Moth Dataset contains 675 species of Geometridae moths quantitatively collected from tropical montane rainforests in southern Ecuador. This dataset is extremely challenging, as it includes numerous morphologically similar species that remain scientifically undescribed. The Costa Rica Dataset covers a diverse collection of moths and butterflies extensively sampled across northwestern Costa Rica. This dataset is already publicly available, and transformed metadata will be released to facilitate usage by the computer vision community. Both datasets are designed to address species identification and classification tasks in biodiversity research, enabling automated vision systems to assist experts with species differentiation and annotation, as well as conduct systematic and quantitative analyses of inter-species differences.