Neural Unsigned Distance Fields for Implicit Function Learning

By Julian Chibane, Aymen Mir, Gerard Pons-Moll. Neural Information Processing Systems, 2020

406 citations recorded by the source corpus.

In this work we target a learnable output representation that allows continuous, high resolution outputs of arbitrary shape. Recent works represent 3D surfaces implicitly with a Neural Network, thereby breaking previous barriers in resolution, and ability to represent diverse topologies. However, neural implicit representations are limited to closed surfaces, which divide the space into inside and outside. Many real world objects such as walls of a scene scanned by a sensor, clothing, or a car with inner structures are not closed. This constitutes a significant barrier, in terms of data pre-processing (objects need to be artificially closed creating artifacts), and the ability to output open surfaces. In this work, we propose Neural Distance Fields (NDF), a neural network based model which predicts the unsigned distance field for arbitrary 3D shapes given sparse point clouds. NDF represent surfaces at high resolutions as prior implicit models, but do not require closed surface data, and significantly broaden the class of representable shapes in the output. NDF allow to extract the surface as very dense point clouds and as meshes. We also show that NDF allow for surface normal calculation and can be rendered using a slight modification of sphere tracing. We find NDF can be used for multi-target regression (multiple outputs for one input) with techniques that have been exclusively used for rendering in graphics. Experiments on ShapeNet show that NDF, while simple, is the state-of-the art, and allows to reconstruct shapes with inner structures, such as the chairs inside a bus. Notably, we show that NDF are not restricted to 3D shapes, and can approximate more general open surfaces such as curves, manifolds, and functions. Code is available for research at this https URL.

Subfields: Computational Mechanics, Computer Graphics and Computer-Aided Design

View Paper: https://www.semanticscholar.org/paper/5336c8aab920ff9c29bbdb5fbab83fad44b2276e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Computer Graphics and Computer-Aided Design

Citations in the corpus

BEHAVE: Dataset and Method for Tracking Human Object Interactions (2022, methodology): “4.1) and use this to predict the unsigned distance [18] to human and object surface u oj , u hj = f udf φ ( F j ) .”
Deep Marching Tetrahedra: a Hybrid Representation for High-Resolution 3D Shape Synthesis (2021, neutral): “performance in many applications, including single view 3D reconstruction [60, 30, 47, 48], shape manipulation, and synthesis [26, 21, 28, 14, 1, 9].”
D-NeRF: Neural Radiance Fields for Dynamic Scenes (2020, background): “Recently, there has been a strong burst in representing 3D data in an implicit manner via a neural network [29, 36, 7, 55, 9, 11, 16].”
What's Behind the Couch? Directed Ray Distance Functions (DRDF) for 3D Scene Reconstruction (2021, positive): “Recent works have demonstrated overfitting of neural networks to single scenes [1, 2, 9, 54] but none attempt to predict scene-level 3D conditioned on an image.”
Revealing Occlusions with 4D Neural Fields (2022, background): “Neural implicit functions have become very popular for 3D representation in recent years [12, 48, 50, 67], building on the seminal work of Neural Radiance Fields (NeRF) [34] and neural implicit surface modeling [39].”
StyleSDF: High-Resolution 3D-Consistent Image and Geometry Generation (2021, background): “Neural implicit representations (NIR) have been widely adopted for learning shape and appearance of objects [4,10, 14, 20, 45, 50, 52, 56, 57], local parts [17, 18], and full 3D scenes [7, 11, 29, 55] from explicit 3D supervisions.”
Fast end-to-end learning on protein surfaces (2020, background): “based on 2D views [44], volumetric [38], space partitioning [35, 42, 39] and implicit representations [14]), geometric deep learning seeks to develop a non-Euclidean analogy of filtering and pooling operations.”
iMAP: Implicit Mapping and Positioning in Real-Time (2021, background): “Implicit Scene Representation with MLPs Scene representation and graphics have seen much recent progress on using implicit MLP neural models for object reconstruction [19, 14], object compression [33] novel view synthesis [15], and scene completion [27, 3].”