3 D ShapeNets : A Deep Representation for Volumetric Shape Modeling

By Zhirong Wu, Shuran Song, Aditya Khosla, Fisher Yu, Linguang Zhang, Xiaoou Tang, Jianxiong Xiao. 2015

How researchers see this paper

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

Citations in the corpus

Multi-view Convolutional Neural Networks for 3D Shape Recognition (2015, methodology): “For our experiments, we use the same training and test split of ModelNet40 as in [37].2 Our shape descriptors are compared against the 3D ShapeNets by Wu et al. [37], the Spherical Harmonics descriptor (SPH) by Kazhdan et al. [16], the LightField descriptor (LFD) by Chen et al. [5], and Fisher vecto”
Generative and Discriminative Voxel Modeling with Convolutional Neural Networks (2016, background): “We also note that our highest performance on ModelNet10 came from models trained on ModelNet40, despite ModelNet40 not containing any additional ModelNet10 instances.”
Category-specific object reconstruction from a single image (2014, neutral): “More expressive non-linear shape models [38] may prove helpful, as well as a tighter integration between segmentation and reconstruction.”
PRNet: Self-Supervised Learning for Partial-to-Partial Registration (2019, unknown): “We evaluate partial-to-partial registration on ModelNet40 [62].”
Geodesic Convolutional Neural Networks on Riemannian Manifolds (2015, methodology): “CNNs have been applied to 3D data in the very recent works [48, 44] using standard (Euclidean) CNN architectures applied to volumetric 2D views shape representations, making them unsuitable for deformable shapes.”
Dreaming More Data: Class-dependent Distributions over Diffeomorphisms for Learned Data Augmentation (2015, background): “As deep models are beginning to appear for three-dimensional data [49] it would be interesting to combine them with learned augmentation schemes.”
FAST SEMANTIC SEGMENTATION OF 3D POINT CLOUDS WITH STRONGLY VARYING DENSITY (2016, methodology): “To our knowledge, it has only been used – with some success – for 3D voxel grids (Lai et al., 2014, Wu et al., 2015), but not yet for generic point clouds; Presumably because the convolutional strategy is not efﬁcient in the absence of a regular neighborhood structure.”
Learning Dense Correspondence via 3D-Guided Cycle Consistency (2016, methodology): “In cases where depth data is available, deep learning methods have recently been applied to 3D object recognition and alignment between CAD models and RGB-D images [13, 33, 37].”
O-CNN (2017, methodology): “Wu et al .”
SUN RGB-D: A RGB-D scene understanding benchmark suite (2015, background): “There are many interesting works on RGB-D scene understanding , including semantic segmentation [53] [49] [19] object classification [69], object detection [59] [20] [62], context reasoning [38], mid-level recognition [32, 31], and surface orientation and room layout estimation [13, 14, 74].”
GIFT: A Real-Time and Scalable 3D Shape Search Engine (2016, background): “ModelNet is a large-scale 3D CAD model dataset introduced by Wu et al .”
Deep Closest Point: Learning Representations for Point Cloud Registration (2019, methodology): “1 on ModelNet40 [52], using DCP-v1 and DCP-v2.”
Enhanced Deep‐Learning Prediction of Molecular Properties via Augmentation of Bond Topology (2018, methodology): “Employing the recognition power of CNN, voxel-based models were introduced in molecular interpretation by rasterizing a 3D molecular structure into angstrom-divided voxels to interpret molecules as an image.(20,21) This”