Corpus ID: 8785126

Submanifold Sparse Convolutional Networks

@article{Graham2017SubmanifoldSC,
  title={Submanifold Sparse Convolutional Networks},
  author={Benjamin Graham and Laurens van der Maaten},
  journal={ArXiv},
  year={2017},
  volume={abs/1706.01307}
}
Convolutional network are the de-facto standard for analysing spatio-temporal data such as images, videos, 3D shapes, etc. [...] Key Method We introduce a sparse convolutional operation tailored to processing sparse data that differs from prior work on sparse convolutional networks in that it operates strictly on submanifolds, rather than "dilating" the observation with every layer in the network. Our empirical analysis of the resulting submanifold sparse convolutional networks shows that they perform on par…Expand
3D Semantic Segmentation with Submanifold Sparse Convolutional Networks
TLDR
This work introduces new sparse convolutional operations that are designed to process spatially-sparse data more efficiently, and uses them to develop Spatially-Sparse Convolutional networks, which outperform all prior state-of-the-art models on two tasks involving semantic segmentation of 3D point clouds. Expand
SBNet: Sparse Blocks Network for Fast Inference
TLDR
This work leverages the sparsity structure of computation masks and proposes a novel tiling-based sparse convolution algorithm that is effective on LiDAR-based 3D object detection, and reports significant wall-clock speed-ups compared to dense convolution without noticeable loss of accuracy. Expand
SPLATNet: Sparse Lattice Networks for Point Cloud Processing
  • Hang Su, V. Jampani, +4 authors J. Kautz
  • Computer Science
  • 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition
  • 2018
TLDR
A network architecture for processing point clouds that directly operates on a collection of points represented as a sparse set of samples in a high-dimensional lattice that outperforms existing state-of-the-art techniques on 3D segmentation tasks. Expand
Generative Sparse Detection Networks for 3D Single-shot Object Detection
TLDR
Generative Sparse Detection Network is proposed, a fully-convolutional single-shot sparse detection network that efficiently generates the support for object proposals and can process unprecedentedly large-scale inputs with a single fully- Convolutional feed-forward pass. Expand
Sparse and Dense Data with CNNs: Depth Completion and Semantic Segmentation
TLDR
This proposal efficiently learns sparse features without the need of an additional validity mask, and works with densities as low as 0.8% (8 layer lidar). Expand
SECOND: Sparsely Embedded Convolutional Detection
TLDR
An improved sparse convolution method for Voxel-based 3D convolutional networks is investigated, which significantly increases the speed of both training and inference and introduces a new form of angle loss regression to improve the orientation estimation performance. Expand
ConvPoint: continuous convolutions for cloud processing
TLDR
This paper proposes a generalization of the discrete convolutional neural networks able to deal with sparse input point cloud and replaces the discrete kernels by continuous ones, which can easily be used for neural network design similarly to 2D CNNs. Expand
SG-NN: Sparse Generative Neural Networks for Self-Supervised Scene Completion of RGB-D Scans
TLDR
A novel approach that converts partial and noisy RGB-D scans into high-quality 3D scene reconstructions by inferring unobserved scene geometry and combined with a new 3D sparse generative convolutional neural network architecture is able to predict highly detailed surfaces in a coarse-to-fine hierarchical fashion. Expand
Inference, Learning and Attention Mechanisms that Exploit and Preserve Sparsity in CNNs
TLDR
This work introduces a suite of tools that exploit sparsity in both the feature maps and the filter weights of a CNN, and thereby allow for significantly lower memory footprints and computation times than the conventional dense framework, when processing data with a high degree of sparsity. Expand
Quadtree Convolutional Neural Networks
TLDR
The method decomposes and represents the image as a linear quadtree that is only refined in the non-empty portions of the image, enabling QCNN to learn from sparse images much faster and process high resolution images without the memory constraints faced by traditional CNNs. Expand
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 12 REFERENCES
Vote3Deep: Fast object detection in 3D point clouds using efficient convolutional neural networks
TLDR
This paper proposes a computationally efficient approach to detecting objects natively in 3D point clouds using convolutional neural networks (CNNs) by leveraging a feature-centric voting scheme to implement novel convolutionan layers which explicitly exploit the sparsity encountered in the input. Expand
Very Deep Convolutional Networks for Large-Scale Image Recognition
TLDR
This work investigates the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting using an architecture with very small convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. Expand
OctNet: Learning Deep 3D Representations at High Resolutions
TLDR
The utility of the OctNet representation is demonstrated by analyzing the impact of resolution on several 3D tasks including 3D object classification, orientation estimation and point cloud labeling. Expand
Densely Connected Convolutional Networks
TLDR
The Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion, and has several compelling advantages: they alleviate the vanishing-gradient problem, strengthen feature propagation, encourage feature reuse, and substantially reduce the number of parameters. Expand
Aggregated Residual Transformations for Deep Neural Networks
TLDR
On the ImageNet-1K dataset, it is empirically show that even under the restricted condition of maintaining complexity, increasing cardinality is able to improve classification accuracy and is more effective than going deeper or wider when the authors increase the capacity. Expand
Deconvolutional networks
TLDR
This work presents a learning framework where features that capture these mid-level cues spontaneously emerge from image data, based on the convolutional decomposition of images under a spar-sity constraint and is totally unsupervised. Expand
Going deeper with convolutions
We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual RecognitionExpand
Volumetric and Multi-view CNNs for Object Classification on 3D Data
TLDR
This paper introduces two distinct network architectures of volumetric CNNs and examines multi-view CNNs, providing a better understanding of the space of methods available for object classification on 3D data. Expand
Identity Mappings in Deep Residual Networks
TLDR
The propagation formulations behind the residual building blocks suggest that the forward and backward signals can be directly propagated from one block to any other block, when using identity mappings as the skip connections and after-addition activation. Expand
Sparse 3D convolutional neural networks
TLDR
A convolutional neural network designed for processing sparse three-dimensional input data and experiment with CNNs on the 2D triangular-lattice and 3D tetrahedral- lattice is implemented. Expand
...
1
2
...