SketchyCOCO: Image Generation From Freehand Scene Sketches

TLDR

The paper proposes EdgeGAN, the first method for automatic image generation from scene‑level freehand sketches. EdgeGAN uses an attribute‑vector bridged GAN trained on the newly created SketchyCOCO dataset to generate images controllably from freehand sketches at both object and scene levels. Experiments show that EdgeGAN produces realistic, complex scene images from diverse freehand sketches, as confirmed by quantitative metrics, human judgments, and ablation studies.

Abstract

We introduce the first method for automatic image generation from scene-level freehand sketches. Our model allows for controllable image generation by specifying the synthesis goal via freehand sketches. The key contribution is an attribute vector bridged Generative Adversarial Network called EdgeGAN, which supports high visual-quality object-level image content generation without using freehand sketches as training data. We have built a large-scale composite dataset called SketchyCOCO to support and evaluate the solution. We validate our approach on the tasks of both object-level and scene-level image generation on SketchyCOCO. Through quantitative, qualitative results, human evaluation and ablation studies, we demonstrate the method's capacity to generate realistic complex scene-level images from various freehand sketches.