Hyperspectral imaging generated adulterated honey dataset

This is a generated dataset of hyperspectral images of adulterated honey. The dataset was generated using a variational autoencoder and a latent space transform based on the ground truth honey adulteration data. The generated data utilises the larger dataset of botanical origins honey, and covers the same brands and classes of honey at each adulteration concentration (0%, 5%, 10%, 25%, 50%). This is a generated dataset, so is not strictly real ground-truth data, however the trends followed in the generated data are similar to the ground-truth adulteration. This dataset presents a reasonable approximation of a larger dataset of adulterated honey.

本数据集为掺假蜂蜜（adulterated honey）高光谱图像（hyperspectral images）的生成数据集。本数据集基于真实掺假蜂蜜真值（Ground Truth）数据，通过变分自编码器（Variational Autoencoder）与隐空间变换（latent space transform）生成。该生成数据依托规模更大的植物源蜂蜜数据集构建，在0%、5%、10%、25%、50%各掺假浓度梯度下，均覆盖了同款蜂蜜的品牌与品类。本数据集为生成式数据集，并非严格意义上的真实真值数据，但其数据变化趋势与真实掺假蜂蜜的真值数据趋势高度一致。本数据集可合理近似规模更大的掺假蜂蜜数据集。