ControlSketch Dataset
收藏资源简介:
ControlSketch数据集是由Tel Aviv University等机构创建的合成图像-向量草图对数据集,包含超过35,000对高质量图像和对应的向量草图,跨越100个类别。数据集中的图像使用SDXL生成,草图则通过ControlSketch优化技术产生。该数据集旨在训练能够高效生成图像向量草图的生成模型,解决现有草图数据集存在的专业质量不足和规模有限的问题。
The ControlSketch dataset is a synthetic image-vector sketch pair dataset created by institutions including Tel Aviv University and other relevant organizations. It contains over 35,000 high-quality pairs of images and their corresponding vector sketches, spanning 100 distinct categories. The images in the dataset are generated using SDXL, while the vector sketches are produced via the ControlSketch optimization technique. This dataset is designed to train generative models that can efficiently generate vector sketches from images, addressing the issues of insufficient professional quality and limited scale in existing sketch datasets.

- 1SwiftSketch: A Diffusion Model for Image-to-Vector Sketch GenerationTel Aviv University, Reichman University, MIT · 2025年



