INTRODUCTION:
Deep Face Drawing allows users with little to no training in drawing to produce high-quality face images from rough or even incomplete freehand sketches.
The fast generation of face images from freehand sketches is now possible using image-to-image deep translation techniques. The project first aims to learn feature embeddings of key face components and push corresponding parts of input sketches towards underlying component manifolds defined by the feature vectors of face component samples.
The framework consists of three main modules and trained in two-stage.
1) CE (Component Embedding)
2) FM (Feature Mapping)
3) IS (Image Synthesis)
How To Generate Realistic Faces From Sketches?
The left-hand side consists of the sketches in black and white and the right-hand side is of course, the reconstruction of the sketches. Now, the sketches themselves are admittedly more detailed than average but even then, the fact that these conversions are 3d face reconstruction from a single image is brilliant. Some of the photos in this gallery have a bit of extra shading on the jawline, hair, eyes, and various parts of the face.
The algorithm manages to catch onto those details as well and it clearly reflects in the conversion. The bottom photo of the second row, for instance, has a black shade drawn on the jawline as it is in the sketch. This algorithm also has the ability to come up with faces that don’t even exist.
Artists and people without any training or experience in making sketches can benefit from this project for creating face images. The tool is easy to use even for non-artists, while still supporting fine-grained control of shape details.