Learning to Approximate Directional Fields Defined over 2D Planes
This work addresses the need for efficient and transferable directional field reconstruction in geometry processing, but it appears incremental as it applies deep learning to an existing problem without claiming major breakthroughs.
The paper tackles the problem of reconstructing directional fields for geometry processing tasks by proposing a deep learning-based approach to overcome the limitations of complex optimization procedures, and it studies the expressive power and generalization ability of this method.
Reconstruction of directional fields is a need in many geometry processing tasks, such as image tracing, extraction of 3D geometric features, and finding principal surface directions. A common approach to the construction of directional fields from data relies on complex optimization procedures, which are usually poorly formalizable, require a considerable computational effort, and do not transfer across applications. In this work, we propose a deep learning-based approach and study the expressive power and generalization ability.