3d Shapes
PulseAugur coverage of 3d Shapes — every cluster mentioning 3d Shapes across labs, papers, and developer communities, ranked by signal.
-
New research questions linear probe methodology in vision models
A new paper questions the common methodology of using linear probes to determine if pretrained vision models encode generative factors linearly. The research suggests that high probe accuracy can be achieved by models w…
-
New descriptor precisely analyzes 3D shapes, preserving chirality
Researchers have developed a new descriptor for analyzing 3D shapes with spherical topology, addressing limitations of existing methods. This descriptor precisely removes dependencies on parameterization, pose, and scal…
-
Gromov-Wasserstein Quantization Extends K-Means for Geometry-Aware Clustering
This paper introduces Gromov-Wasserstein (GW) quantization as an extension of traditional k-means clustering. Unlike standard Wasserstein quantization which clusters points within a space, GW quantization also considers…
-
New method links image segments to 3D shape correspondences
Researchers have developed a new method for establishing correspondences between image segments and 3D shapes, addressing challenges posed by differences in appearance, geometry, and viewpoint. The approach distills dee…
-
SGSoft pipeline learns 3D shape correspondences via template-guided signals
Researchers have introduced SGSoft, a new pipeline for establishing dense correspondences across deformable 3D shapes. This method uses a canonical template to create a geodesic correspondence field, which then guides t…
-
SQuadGen uses diffusion models and chart distance fields for simple 3D quad mesh layouts
Researchers have developed SQuadGen, a new diffusion-based generative framework designed to create simple quad mesh layouts on 3D shapes. This method addresses the challenge of generating clean, artist-friendly quad lay…