HLSS/SAM_Preprocessed_MHB

SAM_Preprocesed_MHB是MassHumanBurns数据集的一个变体，所有掩码均使用``sam2.1_hiera_tiny``重新划分。该数据集包含模拟烧伤模式的人体图像掩码，每个样本包括前后视图、全身掩码、烧伤区域掩码以及烧伤百分比标签。数据集由8种不同的人体模型变体组成，每种变体有625个不同的人体网格，总计5,000个人体网格。训练样本包括每种变体的前500个人体网格（总计4,000 x 8），测试样本包括每种变体的最后125个人体网格（总计1,000 x 8）。每个烧伤样本有2个视图（前后）和4个不同的相机角度。烧伤级别为每个人体模型的8个严重程度级别。数据集目录结构包括burn_masks、full_masks、gender.json和labels.npy等文件。

SAM_Preprocesed_MHB is a variant of the MassHumanBurns dataset. All the masks have been re-divided using ``sam2.1_hiera_tiny``. This dataset consists of corresponding masks of human figures with simulated burn patterns. Each sample includes front and back views, full body masks, and burn region masks, along with ground truth burn percentage labels. The dataset comprises 8 variations of human models, each with 625 different human meshes, totaling 5,000 human meshes. Training samples include the first 500 human meshes per variation (4,000 x 8 total), and test samples include the last 125 human meshes per variation (1,000 x 8 total). Each burn sample has 2 views (front/back) and 4 distinct camera angles. Burn levels are 8 severity levels per human model. The dataset directory structure includes burn_masks, full_masks, gender.json, and labels.npy files.