project-oceania/uvp6net

UVP6Net数据集包含了使用UVP6设备捕捉到的浮游生物图像。数据集中的图像经过UVP6固件的预处理，记录了每个单独对象的感兴趣区域（ROIs）。数据集的特征包括图像和标签，标签代表110个不同的类别。为了训练机器学习分类器，图像被分为训练集、验证集和测试集。此外，还提供了一个额外的图像子集，用于训练实际部署在NKE CTS5浮标上的UVP6嵌入式分类模型，该模型的类别数量从110个减少到20个，以确保在功率受限的硬件上具有足够的性能。该数据集在CC BY-NC 4.0许可下提供。

The UVP6Net dataset includes plankton images captured with the UVP6 device. The images are processed by the UVP6 firmware, with recorded regions of interest (ROIs) around individual objects. The dataset features include images and labels, with the labels representing 110 different classes. The images are split into training, validation, and test sets for machine learning classifier training. An additional subset of images is provided for training an embedded classification model deployed on NKE CTS5 floats, with a coarser classification of 20 classes to ensure adequate performance on power-constrained hardware. The dataset is licensed under CC BY-NC 4.0.