pytc/DenSpineEM

DenSpine数据集包含来自3个大脑样本的树突：`seg_den`（也称为`M50`）、`mouse`（`M10`）和`human`（`H10`）。每个样本都有3个体积的`.h5`文件，分别是：原始树突的实例分割、二进制分割，以及树突脊柱的实例分割。此外，还提供了沿树突中心线骨架采样的预处理点云文件，以方便评估基于点云的方法。数据集还提供了`seg_den`的训练划分。

The DenSpine dataset consists of dendrites from 3 brain samples: `seg_den` (also known as `M50`), `mouse` (`M10`), and `human` (`H10`). Each species has 3 volumetric `.h5` files: the raw segmentation of entire dendrites, a binary segmentation, and spine instance segmentation. In addition, preprocessed point clouds sampled along the centerline skeletons of dendrites are provided for evaluating point-cloud based methods. The dataset also includes training splits for the `seg_den` dataset.