SketchyCOCO

SketchyCOCO数据集由两部分组成:
对象级数据
对象级数据包含 $20198(train18869 val1329)$ 三重 {前景草图，前景图像，前景边缘图} 示例，涵盖14个类，$27683(train22171 val5512)$ {背景草图，背景图像} 示例，涵盖3个类。
场景级数据
场景级数据包含 $14081 (火车11265 val 2816)$ 对 {前景图像和背景草图，场景图像} 示例，$14081 (火车11265 val 2816)$ 对 {场景草图，场景图像} 示例和 $14081 (火车11265 val 2816)$ 场景草图的分割地面真相。一些val场景图像来自COCO-Stuff数据集的火车图像，用于增加SketchyCOCO数据集的val图像的数量。

The SketchyCOCO dataset consists of two parts:
Object-level data
Object-level data contains 20198 (train: 18869, val: 1329) triplets of {foreground sketch, foreground image, foreground edge map}, covering 14 categories, and 27683 (train: 22171, val: 5512) pairs of {background sketch, background image}, covering 3 categories.
Scene-level data
Scene-level data contains 14081 (train: 11265, val: 2816) pairs of {foreground image and background sketch, scene image}, 14081 (train: 11265, val: 2816) pairs of {scene sketch, scene image}, and 14081 (train: 11265, val: 2816) segmentation ground truths for scene sketches. Some validation scene images are sourced from the training images of the COCO-Stuff dataset to increase the number of validation images in the SketchyCOCO dataset.

搜集汇总

背景与挑战

背景概述

SketchyCOCO是一个用于草图到图像翻译的数据集，包含对象级和场景级数据，涵盖多种类别的草图、图像及边缘图。该数据集由华为诺亚方舟实验室、中山大学和南京大学联合发布，适用于计算机视觉领域的研究。