Deep Implicit Surface Point Prediction Networks

Venkatesh, R.; Karmali, Tejan; Sharma, Shubham; Ghosh, Aurobrata; Babu, R. Venkatesh; Jeni, László A.; Singh, Maneesh

doi:10.1109/iccv48922.2021.01242

Cited by 36 publications

(20 citation statements)

References 36 publications

(73 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We do the same for our method by meshing our learned current using marching cubes. Our method consistently achieves better quality reconstructions than [59].…”

Section: Surface Reconstructionmentioning

confidence: 86%

“…Because neural implicit shapes are typically level sets of learned functions, this limits the class of representable shapes to closed surfaces. Two notable exceptions are [9], which learns unsigned distance functions rather than SDFs, and [59], which maps an input point to its closest point on the target surface. Our DeepCurrents adopt a hybrid representation, which models boundaries explicitly and allows them to be used as handles for manipulation.…”

Section: Deep Learning For Shape Reconstructionmentioning

confidence: 99%

“…Our currents faithfully reconstruct the target geometry. Additionally, we compare quantitatively to [59] in Table 1. We train their model for the same amount of time as ours on randomly picked models.…”

Section: Surface Reconstructionmentioning

confidence: 99%

“…Model UCD ( [59]) UCD (ours) head 0.0049 0.0010 hand 0.0045 0.0011 torso 0.0049 0.00092 foot 0.0055 0.00092 Table 1. Quantitative comparison of unidirectional Chamfer distance to [59] on single surface reconstruction of random models from each each of four shape categories.…”

Section: Surface Reconstructionmentioning

confidence: 99%

See 3 more Smart Citations

DeepCurrents: Learning Implicit Representations of Shapes with Boundaries

Palmer¹,

Smirnov²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Recent techniques have been successful in reconstructing surfaces as level sets of learned functions (such as signed distance fields) parameterized by deep neural networks. Many of these methods, however, learn only closed surfaces and are unable to reconstruct shapes with boundary curves. We propose a hybrid shape representation that combines explicit boundary curves with implicit learned interiors. Using machinery from geometric measure theory, we parameterize currents using deep networks and use stochastic gradient descent to solve a minimal surface problem. By modifying the metric according to target geometry coming, e.g., from a mesh or point cloud, we can use this approach to represent arbitrary surfaces, learning implicitly defined shapes with explicitly defined boundary curves. We further demonstrate learning families of shapes jointly parameterized by boundary curves and latent codes.

show abstract

“…We do the same for our method by meshing our learned current using marching cubes. Our method consistently achieves better quality reconstructions than [59].…”

Section: Surface Reconstructionmentioning

confidence: 86%

Section: Deep Learning For Shape Reconstructionmentioning

confidence: 99%

Section: Surface Reconstructionmentioning

confidence: 99%

Section: Surface Reconstructionmentioning

confidence: 99%

See 2 more Smart Citations

DeepCurrents: Learning Implicit Representations of Shapes with Boundaries

Palmer¹,

Smirnov²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Inferring the 3D shape from a single image is an inherently ill-posed task -an image of a chair from the back does not remove ambiguities about the shape of its seat. Several approaches have shown impressive single-view reconstruction results using voxels [10,14,31,38,39], point clouds [13,19,41], meshes [36,37], and most recently implicit representations of 3D surfaces like SDFs [17,43], UDFs [9] and CSPs [35]. However, these are often deterministic in nature and only generate a 3D single output.…”

Section: Related Workmentioning

confidence: 99%

AutoSDF: Shape Priors for 3D Completion, Reconstruction and Generation

Mittal¹,

Cheng²,

Singh³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Few ‘Zero Level Set’-Shot Learning of Shape Signed Distance Functions in Feature Space

Ouasfi

Boukhayma

2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

While current state-of-the-art generalizable implicit neural shape models [7,54] rely on the inductive bias of convolutions, it is still not entirely clear how properties emerging from such biases are compatible with the task of 3D reconstruction from point cloud. We explore an alternative approach to generalizability in this context. We relax the intrinsic model bias (i.e. using MLPs to encode local features as opposed to convolutions) and constrain the hypothesis space instead with an auxiliary regularization related to the reconstruction task, i.e. denoising. The resulting model is the first only-MLP locally conditioned implicit shape reconstruction from point cloud network with fast feed forward inference. Point cloud borne features and denoising offsets are predicted from an exclusively MLP-made network in a single forward pass. A decoder predicts occupancy probabilities for queries anywhere in space by pooling nearby features from the point cloud order-invariantly, guided by denoised relative positional encoding. We outperform the state-of-the-art convolutional method [7] while using half the number of model parameters.

show abstract

Deep Implicit Surface Point Prediction Networks

Cited by 36 publications

References 36 publications

DeepCurrents: Learning Implicit Representations of Shapes with Boundaries

DeepCurrents: Learning Implicit Representations of Shapes with Boundaries

AutoSDF: Shape Priors for 3D Completion, Reconstruction and Generation

Few ‘Zero Level Set’-Shot Learning of Shape Signed Distance Functions in Feature Space

Contact Info

Product

Resources

About