O-CNN

By Peng-Shuai Wang, Yang Liu, Yu-Xiao Guo, Chun-Yu Sun, Xin Tong. 2017

194 citations recorded by the source corpus.

We present O-CNN, an Octree-based Convolutional Neural Network (CNN) for 3D shape analysis. Built upon the octree representation of 3D shapes, our method takes the average normal vectors of a 3D model sampled in the finest leaf octants as input and performs 3D CNN operations on the octants occupied by the 3D shape surface. We design a novel octree data structure to efficiently store the octant information and CNN features into the graphics memory and execute the entire O-CNN training and evaluation on the GPU. O-CNN supports various CNN structures and works for 3D shapes in different representations. By restraining the computations on the octants occupied by 3D surfaces, the memory and computational costs of the O-CNN grow quadratically as the depth of the octree increases, which makes the 3D CNN feasible for high-resolution 3D models. We compare the performance of the O-CNN with other existing 3D CNN solutions and demonstrate the efficiency and efficacy of O-CNN in three shape analysis tasks, including object classification, shape retrieval, and shape segmentation.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition, Model Architectures

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 Vision and Pattern Recognition

In Model Architectures

Citations in the corpus

3D Instance Segmentation of MVS Buildings (2021, methodology): “Volumetric methods ease this issue using a 3D grid representation, which is notoriously expensive in terms of computational efficiency and memory consumption [4], [12], [36], [37], [38], [39].”
PointASNL: Robust Point Clouds Processing Using Nonlocal Neural Networks With Adaptive Sampling (2020, methodology): “Various subsequent improvement methods have been proposed, e.g., efficient spatial-temporal convolution MinkowskiNet [5], computational effective Submanifold sparse convolution [7], and Oc-tree based neural networks O-CNN [39] and OctNet [27].”
Multi-directional geodesic neural networks via equivariant convolution (2018, methodology): “Due to the potential memory complexity of this approach, several efficient extensions have been proposed, e.g., [Klokov and Lempitsky 2017; Wang et al. 2017], and a comparison between view-based and volumetric approaches has been presented in [Qi et al. 2016].”
Rotation Invariant Convolutions for 3D Point Clouds Deep Learning (2019, methodology): “Some methods exploit hierarchical structures and clustering for scalability [28, 15, 33, 34, 16], mapping point cloud to two dimensional space [40, 9, 18], applying spectral analysis [42], or addressing non-uniform point distribution [10].”
Deep Learning 3D Shapes Using Alt-az Anisotropic 2-Sphere Convolution (2018, methodology): “A straightforward extension is to treat a 3D shape as a voxel grid (Wu et al. (2015); Maturana & Scherer (2015); Song & Xiao (2016); Wang et al. (2017); Riegler et al. (2016).)”
PartNet: A Large-Scale Benchmark for Fine-Grained and Hierarchical Part-Level 3D Object Understanding (2018, background): “Recent advances of 3D semantic segmentation [30, 31, 46, 19, 35, 24, 9, 39, 40, 42, 33, 7, 26, 23] have accomplished promising performance in coarse-level segmentation on the ShapeNet Part dataset [3, 45].”
Mask3D: Mask Transformer for 3D Semantic Instance Segmentation (2022, methodology): “Our model, called Mask3D, builds on recent advances in both Transformers [5, 37] and 3D deep learning [8, 17, 57]: to compute strong point features, we leverage a sparse convolutional feature backbone [8] that efficiently”
Relation-Shape Convolutional Neural Network for Point Cloud Analysis (2019, methodology): “Recent space partition methods like K-d trees [16] or octrees [39, 36, 28] rescue some resolution issues but still rely on the subdivision of a bounding volume rather than a local geometric shape.”
Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matching (2019, background): “Works in [28, 28, 37] partition advanced space to reduce memory consumption and construct a fixed cost volume representation which lacks flexibility.”
PointGrid: A Deep Network for 3D Shape Understanding (2018, methodology): “Comparing with methods applied to point cloud, despite of its simplicity, our network edges out Set-Conv [41], PointNet [38], PointNet++ [40], Kd-Net [27] and O-CNN [55].”
SPLATNet: Sparse Lattice Networks for Point Cloud Processing (2018, methodology): “Instead of using fixed grids, more recent approaches preprocess the input shapes into adaptively subdivided, hierarchical grids with denser cells placed near the surface [35, 34, 25, 42, 40].”
Recurrent MVSNet for High-Resolution Multi-View Stereo Depth Inference (2019, methodology): “OctNet [23] and O-CNN [27] exploit the sparsity in 3D data and introduce the octree structure to 3D CNNs.”