MassID45: Mixed Arthropod Sample Segmentation and Identification

A dataset for training automatic classifiers of bulk insect samples, using both molecular and imaging data. DNA barcodes and images are available at both the unsorted sample level and the full set of individual specimens from those samples. Further information about the dataset is available in the pre-print: A multi-modal dataset for insect biodiversity with imagery and DNA at the trap and individual level. Code for preprocessing and training/inference is available at https://github.com/uoguelph-mlrg/MassID45.本数据集旨在训练针对批量昆虫样本的自动分类器，同时融合分子数据与成像数据两类模态。数据集同时提供未分拣样本层级，以及对应样本的全部单个标本层级的DNA条形码（DNA barcode）与图像数据。 有关本数据集的更多详细信息可参阅预印本论文《兼具诱捕器级与个体级影像及DNA数据的昆虫生物多样性多模态数据集》。数据集预处理、模型训练与推理的相关代码可通过以下链接获取：https://github.com/uoguelph-mlrg/MassID45。