Dilated convolutions, also known as atrous convolutions, have been widely
explored in deep convolutional neural networks (DCNNs) for various tasks like
semantic image segmentation, object detection, audio generation, video
modeling, and machine translation. However, dilated convolutions suffer from
the gridding artifacts, which hampers the performance of DCNNs with dilated
convolutions. In this work, we propose two simple yet effective degridding
methods by studying a decomposition of dilated convolutions. Unlike existing
models, which explore solutions by focusing on a block of cascaded dilated
convolutional layers, our methods address the gridding artifacts by smoothing
the dilated convolution itself. By analyzing them in both the original
operation and the decomposition views, we further point out that the two
degridding approaches are intrinsically related and define separable and shared
(SS) operations, which generalize the proposed methods. We evaluate our methods
thoroughly on two datasets and visualize the smoothing effect through effective
receptive field analysis. Experimental results show that our methods yield
significant and consistent improvements on the performance of DCNNs with
dilated convolutions, while adding negligible amounts of extra training
parameters.
%0 Journal Article
%1 citeulike:14642092
%A Wang, Zhengyang
%A Ji, Shuiwang
%B Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining - KDD '18
%D 2018
%I ACM Press
%K arch backbone dilated
%P 2486--2495
%R 10.1145/3219819.3219944
%T Smoothed Dilated Convolutions for Improved Dense Prediction
%U http://dx.doi.org/10.1145/3219819.3219944
%X Dilated convolutions, also known as atrous convolutions, have been widely
explored in deep convolutional neural networks (DCNNs) for various tasks like
semantic image segmentation, object detection, audio generation, video
modeling, and machine translation. However, dilated convolutions suffer from
the gridding artifacts, which hampers the performance of DCNNs with dilated
convolutions. In this work, we propose two simple yet effective degridding
methods by studying a decomposition of dilated convolutions. Unlike existing
models, which explore solutions by focusing on a block of cascaded dilated
convolutional layers, our methods address the gridding artifacts by smoothing
the dilated convolution itself. By analyzing them in both the original
operation and the decomposition views, we further point out that the two
degridding approaches are intrinsically related and define separable and shared
(SS) operations, which generalize the proposed methods. We evaluate our methods
thoroughly on two datasets and visualize the smoothing effect through effective
receptive field analysis. Experimental results show that our methods yield
significant and consistent improvements on the performance of DCNNs with
dilated convolutions, while adding negligible amounts of extra training
parameters.
%@ 9781450355520
@article{citeulike:14642092,
abstract = {{Dilated convolutions, also known as atrous convolutions, have been widely
explored in deep convolutional neural networks (DCNNs) for various tasks like
semantic image segmentation, object detection, audio generation, video
modeling, and machine translation. However, dilated convolutions suffer from
the gridding artifacts, which hampers the performance of DCNNs with dilated
convolutions. In this work, we propose two simple yet effective degridding
methods by studying a decomposition of dilated convolutions. Unlike existing
models, which explore solutions by focusing on a block of cascaded dilated
convolutional layers, our methods address the gridding artifacts by smoothing
the dilated convolution itself. By analyzing them in both the original
operation and the decomposition views, we further point out that the two
degridding approaches are intrinsically related and define separable and shared
(SS) operations, which generalize the proposed methods. We evaluate our methods
thoroughly on two datasets and visualize the smoothing effect through effective
receptive field analysis. Experimental results show that our methods yield
significant and consistent improvements on the performance of DCNNs with
dilated convolutions, while adding negligible amounts of extra training
parameters.}},
added-at = {2019-02-27T22:23:29.000+0100},
archiveprefix = {arXiv},
author = {Wang, Zhengyang and Ji, Shuiwang},
biburl = {https://www.bibsonomy.org/bibtex/2c098f00fbb0a457fedc3bcd5a4613321/nmatsuk},
booktitle = {Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery \& Data Mining - KDD '18},
citeulike-article-id = {14642092},
citeulike-linkout-0 = {http://arxiv.org/abs/1808.08931},
citeulike-linkout-1 = {http://arxiv.org/pdf/1808.08931},
citeulike-linkout-2 = {http://dx.doi.org/10.1145/3219819.3219944},
day = 27,
doi = {10.1145/3219819.3219944},
eprint = {1808.08931},
interhash = {ce85f75118893c8c85c7e98c4b823d69},
intrahash = {c098f00fbb0a457fedc3bcd5a4613321},
isbn = {9781450355520},
keywords = {arch backbone dilated},
location = {London, United Kingdom},
month = aug,
pages = {2486--2495},
posted-at = {2018-10-02 18:40:16},
priority = {2},
publisher = {ACM Press},
timestamp = {2019-02-27T22:23:29.000+0100},
title = {{Smoothed Dilated Convolutions for Improved Dense Prediction}},
url = {http://dx.doi.org/10.1145/3219819.3219944},
year = 2018
}