SketchyCOCO: Image Generation from Freehand Scene Sketches
This addresses the need for controllable image generation from sketches, which is incremental as it builds on prior sketch-to-image work but extends it to scene-level synthesis.
The authors tackled the problem of generating realistic images from freehand scene sketches, introducing a method that achieves high visual-quality object-level and scene-level image generation without using sketches as training data, as validated on the SketchyCOCO dataset.
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.