Neural Fields as Learnable Kernels for 3D Reconstruction

Williams, Francis H.; Gojčič, Žan; Khamis, Sameh; Zorin, Denis; Bruna, Joan; Fidler, Sanja; Litany, Or

doi:10.48550/arxiv.2111.13674

Cited by 2 publications

(2 citation statements)

References 55 publications

(100 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Neural Shape Modeling Deep-learning methods are now routinely used to model 3D shapes. Most methods rely on auto-encoders or auto-decoders to produce latent vectors that parameterize the target shapes in terms of triangulated meshes [16,32,40], tetrahedral meshes [14,49], surface patches [15], point clouds [1,47], voxel grids [6,11], occupancy functions [8,38,46], signed and unsigned distance fields [9,23], and neural splines [59].…”

Section: Related Workmentioning

confidence: 99%

Deep Active Latent Surfaces for Medical Geometries

Jensen¹,

Wickramasinghe²,

Dahl³

et al. 2022

Preprint

View full text Add to dashboard Cite

Shape priors have long been known to be effective when reconstructing 3D shapes from noisy or incomplete data. When using a deep-learning based shape representation, this often involves learning a latent representation, which can be either in the form of a single global vector or of multiple local ones. The latter allows more flexibility but is prone to overfitting. In this paper, we advocate a hybrid approach representing shapes in terms of 3D meshes with a separate latent vector at each vertex. During training the latent vectors are constrained to have the same value, which avoids overfitting. For inference, the latent vectors are updated independently while imposing spatial regularization constraints. We show that this gives us both flexibility and generalization capabilities, which we demonstrate on several medical image processing tasks. * Equal supervision.Preprint. Under review.

show abstract

Section: Related Workmentioning

confidence: 99%

Deep Active Latent Surfaces for Medical Geometries

Jensen¹,

Wickramasinghe²,

Dahl³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…However, this nonsmoothness may also become more severe when the available data is sparse. While limiting the data may increase the difficulty in generating the correct inductive bias (see Williams et al (2021) for a more in-depth discussion), the manifold embedding may overcome the spurious non-smoothness due to the continuous nature of the hyperplane.…”

Section: Limited Complete Databasementioning

confidence: 99%