academic-datasets/InsectSet459

InsectSet459是一个为自动昆虫识别的机器学习算法开发和测试而设计的综合昆虫声音数据集。它包含了来自459种直翅目（蟋蟀、蝗虫、蝈蝈）和蝉科（蝉）的26,399个音频文件，提供了9.5天的音频材料。数据集覆盖全球范围，重点关注欧洲和北美地区。数据集遵循知识共享授权（CC-BY-4.0或CC0）。数据集分为训练集、验证集和测试集，比例为60/20/20。音频记录来自三个主要来源：xeno-canto、iNaturalist和BioAcoustica。数据整理过程中进行了去重、格式标准化和统一物种命名等操作。

InsectSet459 is a comprehensive dataset of insect sounds designed for developing and testing machine learning algorithms for automatic insect identification. It contains 26,399 audio files from 459 species of Orthoptera (crickets, grasshoppers, katydids) and Cicadidae (cicadas), providing 9.5 days of audio material. The dataset covers worldwide geographic areas with a focus on Europe and North America and is licensed under Creative Commons (CC-BY-4.0 or CC0). The dataset is split into training, validation, and test sets with a 60/20/20 ratio. Audio recordings are sourced from xeno-canto, iNaturalist, and BioAcoustica. The curation process includes deduplication, format standardization, and taxonomic unification.