CO-SKEL dataset

为了评估
广泛的共同骨架化任务，我们开发了一个新的
称为 CO-SKEL 数据集的基准数据集。它由 26 个类别组成，共有 353 张动物图像，
鸟、花和人。这些图像被收集
来自 MSRC 数据集、CosegRep、Weizmann Horses 和
iCoseg 数据集及其真实分割
面具。

To evaluate a wide range of co-skeletonization tasks, we developed a novel benchmark dataset named CO-SKEL. This dataset consists of 26 categories and contains a total of 353 animal images, including birds, flowers and humans. These images are collected from the MSRC Dataset, CosegRep, Weizmann Horses and iCoseg Dataset, along with their ground-truth segmentation masks.

搜集汇总

背景与挑战

背景概述

CO-SKEL dataset是一个用于评估共同骨架化任务的基准数据集，包含26个类别的353张动物图像，涵盖鸟、花和人等类别。这些图像来自多个公开数据集，并附有真实分割掩码。