This respository hosts the datesets for the paper.
Our dataset consists of two parts:
- SketchyScene sentence-level caption
- View in sentence.json
- SketchyScene paragraph-level caption
- View in paragraph.json
Our original image are obtained from the respository SketchyScene
Zou C, Yu Q, Du R, et al. Sketchyscene: Richly-annotated scene sketches[C]//Proceedings of the European Conference on Computer Vision (ECCV). 2018: 421-436.