End-to-End Line Drawing Vectorization

Vector graphics is broadly used in a variety of forms, such as illustrations, logos, posters, billboards, and printed ads. Despite its broad use, many artists still prefer to draw with pen paper, which leads high demand converting raster designs into the vector form. In particular, line drawing primary art attracts research efforts automatically drawings However, existing methods generally adopt two-step approach, stroke segmentation vectorization. Without guidance, raster-based frequently obtains unsatisfying results, over-grouped strokes broken strokes. this we make an attempt proposing end-to-end vectorization method directly generates vectorized primitives from one step. We propose Transformer-based framework perform tracing like human does automatic stroke-by-stroke way novel feature representation multi-modal supervision achieve quality fidelity. Qualitative quantitative evaluations show that our achieves state performance.