Learning Sparse High Dimensional Filters: Image Filtering, Dense CRFs and Bilateral Neural Networks

By Varun Jampani, Martin Kiefel, Peter Gehler. 2015

179 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Mechanics, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Artificial Intelligence

In Computational Mechanics

In Computer Vision and Pattern Recognition

Citations in the corpus

Vote3Deep: Fast object detection in 3D point clouds using efficient convolutional neural networks (2016, methodology): “Another method for performing sparse convolutions is introduced in [12] who make use of “permutohedral lattices”, but only consider comparatively small inputs, as opposed to our work.”
Recurrent Pixel Embedding for Instance Grouping (2017, methodology): “Our recurrent meanshift grouping is reminiscent of other efforts that use unrolled implementations of iterative algorithms such as CRF inference [92] or bilateral filtering [39, 26].”
Deep Joint Image Filtering (2016, background): “The bilateral filters [10,1,4,14,9,15] and guided filters [6] are representative algorithms in this class.”
Self-Supervised Sparse-to-Dense: Self-Supervised Depth Completion from LiDAR and Monocular Camera (2018, background): “As a side note, the inpainting problem also finds close connection to depth denoising [13] and depth super-resolution [14, 15, 16, 17, 18, 19].”
SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences (2019, methodology): “As with voxel-based methods, this scales poorly both in computation and in memory cost and therefore they exploit the sparsity of this representation by using bilateral convolutions [27], which only operates on occupied lattice parts.”
Pyramid Attention Network for Semantic Segmentation (2018, methodology): “0% without MS-COCO [14] and Dense-CRF post-processing[9].”
Survey on semantic segmentation using deep learning techniques (2019, methodology): “Approaches Bilateral Neural Network (BNN)[171] Bilateral filter inference in DenseCRF”
Rethinking Atrous Convolution for Semantic Image Segmentation (2017, background): “Recently, [37] proposes to learn a general and sparse high-dimensional convolution (bilateral convolution), and [73, 7] combine Gaussian Conditional Random Fields and DCNNs for semantic segmentation.”
Pixel-Adaptive Convolutional Neural Networks (2019, methodology): “Existing work [51, 21] back-propagates through the above MF steps to combine CRF inference with CNNs resulting in end-to-end training of CNN-CRF models.”
Adaptive Weighting Multi-Field-Of-View CNN for Semantic Segmentation in Pathology (2019, unknown): “In particular, graphical model [8][22][48][39], spatial pyramid pooling [8][9][47], dilated convolution [46] and multi-scale inputs (i.”
Large Kernel Matters — Improve Semantic Segmentation by Global Convolutional Network (2017, background): “[16] introduces the bilateral filter to learn the specific pairwise potentials within CNN.”
Deep bilateral learning for real-time image enhancement (2017, methodology): “This operation is therefore more expressive than simply applying 3D convolutions in the grid, which would only induce local connectivity on z [Jampani et al. 2016].”
Semantic Segmentation With Context Encoding and Multi-Path Decoding (2020, methodology): “[33] integrated the bilateral solver in DCNN to jointly train them form data.”
RangeNet ++: Fast and Accurate LiDAR Semantic Segmentation (2019, background): “To alleviate this, bilateral convolutions [9] allow them to apply theses operators exclusively on occupied sectors of the lattice.”
O-CNN (2017, background): “It is possible to build similar hierarchical structures based on other lattices, such as the tetrahedral lattice [Graham 2015] and the permutohedral lattice [Jampani et al. 2016], the latter of which could be used for higher-dimension CNNs.”
Group-Free 3D Object Detection via Transformers (2021, background): “Network architecture for Point Cloud A large set of network architectures [38, 12, 29, 9, 47, 23, 44, 39, 45, 28, 30, 33, 43, 40, 17, 1, 41, 49, 10, 22] have been proposed for various point cloud based learning tasks.”