Filters








7,125 Hits in 5.1 sec

Dilated Point Convolutions: On the Receptive Field Size of Point Convolutions on 3D Point Clouds [article]

Francis Engelmann, Theodora Kontogianni, Bastian Leibe
2020 arXiv   pre-print
Our dilated point convolutions alleviate this issue, they significantly increase the receptive field size of point convolutions.  ...  However, we observe that the receptive field size of recent point convolutional networks is inherently limited.  ...  Acknowledgements: This work was supported by the ERC Consolidator Grant DeeViSe(ERC-2017-COG-773161). We thank Mats Steinweg, Dan Jia, Jonas Schult and Alexander Hermans for their valuable feedback.  ... 
arXiv:1907.12046v3 fatcat:57jqxvwdobfxfnx4htwbcvafs4

A Spatiotemporal Dilated Convolutional Generative Network for Point-Of-Interest Recommendation

Chunyang Liu, Jiping Liu, Shenghua Xu, Jian Wang, Chao Liu, Tianyang Chen, Tao Jiang
2020 ISPRS International Journal of Geo-Information  
the user's complicated short- and long-range check-in sequence by using a stack of dilated causal convolution layers and residual block structure.  ...  With the growing popularity of location-based social media applications, point-of-interest (POI) recommendation has become important in recent years.  ...  Conflicts of Interest: The authors declare no conflict of interest.  ... 
doi:10.3390/ijgi9020113 fatcat:4n67u67rc5dt3p7qcj6rnfzkyy

Semantic Segmentation for Point Cloud Scenes via Dilated Graph Feature Aggregation and Pyramid Decoders [article]

Yongqiang Mao, Xian Sun, Wenhui Diao, Kaiqiang Chen, Zonghao Guo, Xiaonan Lu, Kun Fu
2022 arXiv   pre-print
Due to the unicity of receptive field, semantic segmentation of point clouds remains challenging for the expression of multi-receptive field features, which brings about the misclassification of instances  ...  of receptive field bases.  ...  Acknowledgments This work is supported in part by National Natural Science Foundation of China (NSFC) under Grant 61725105.  ... 
arXiv:2204.04944v2 fatcat:3iswkmkgbbfexi4sanoagpqfle

Sequence to Point Learning Based on Bidirectional Dilated Residual Network for Non Intrusive Load Monitoring [article]

Ziyue Jia, Linfeng Yang, Zhenrong Zhang, Hui Liu, Fannie Kong
2020 arXiv   pre-print
To solve these problems, we propose a sequence to point learning framework based on bidirectional (non-casual) dilated convolution for NILM.  ...  The ability of neural networks to extract load features is closely related to its depth.  ...  The authors would like to thank Multi-Function Computer Center of Guangxi University for its high-performance computer.  ... 
arXiv:2006.00250v1 fatcat:tye5xssxbferpj2yiin52kkjom

DGANet: A Dilated Graph Attention-Based Network for Local Feature Extraction on 3D Point Clouds

Jie Wan, Zhong Xie, Yongyang Xu, Ziyin Zeng, Ding Yuan, Qinjun Qiu
2021 Remote Sensing  
Feature extraction on point clouds is an essential task when analyzing and processing point clouds of 3D scenes.  ...  Moreover, by integrating the dilated graph attention module (DGAM) implemented by a novel offset–attention mechanism, the proposed network promises to highlight the differing ability of each edge of the  ...  Conflicts of Interest: The authors declare no conflict of interest.  ... 
doi:10.3390/rs13173484 fatcat:ore6i4n2yncc3dmykjnim7cecy

Log-Polar Space Convolution for Convolutional Neural Networks [article]

Bing Su, Ji-Rong Wen
2021 arXiv   pre-print
The local receptive field grows exponentially with the number of distance levels.  ...  Since the number of parameters increases quadratically with the size of the convolution kernel, many popular models use small convolution kernels, resulting in small local receptive fields in lower layers  ...  Relation with effective receptive field [30] . In [30] , it is found that the effective receptive field only occupies a fraction of the full theoretical receptive field.  ... 
arXiv:2107.11943v1 fatcat:3xtmjmpitjggbge3p324wefqtu

Deformable Convolutional Networks [article]

Jifeng Dai, Haozhi Qi, Yuwen Xiong, Yi Li, Guodong Zhang, Han Hu, Yichen Wei
2017 arXiv   pre-print
Extensive experiments validate the effectiveness of our approach on sophisticated vision tasks of object detection and semantic segmentation. The code would be released.  ...  Both are based on the idea of augmenting the spatial sampling locations in the modules with additional offsets and learning the offsets from target tasks, without additional supervision.  ...  The Aligned-Inception-ResNet model is pre-trained on ImageNet-1K classification [8] . The training procedure follows [22] .  ... 
arXiv:1703.06211v3 fatcat:uc7zirurcbdplamqneuacphnya

Active Convolution: Learning the Shape of Convolution for Image Classification [article]

Yunho Jeon, Junmo Kim
2017 arXiv   pre-print
The convolution layer is the core of the CNN, but few studies have addressed the convolution unit itself. In this paper, we introduce a convolution unit called the active convolution unit (ACU).  ...  Second, the shape of the convolution is learned while training and there is no need to tune it by hand.  ...  VGG [21] is based on the idea that a stack of two convolutional layers with a receptive field 3 × 3 is more effective than a 5 × 5 convolution.  ... 
arXiv:1703.09076v1 fatcat:jxw72xykzfa77alx733aow6lmq

Active Convolution: Learning the Shape of Convolution for Image Classification

Yunho Jeon, Junmo Kim
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
The convolution layer is the core of the CNN, but few studies have addressed the convolution unit itself. In this paper, we introduce a convolution unit called the active convolution unit (ACU).  ...  Second, the shape of the convolution is learned while training and there is no need to tune it by hand.  ...  Acknowledgement This work was partly supported by the ICT R&D program of MSIP/IITP, 2016-0-00563, Research on Adaptive Machine Learning Technology Development for Intelligent Autonomous Digital Companion  ... 
doi:10.1109/cvpr.2017.200 dblp:conf/cvpr/JeonK17 fatcat:emonls2lhrhxrmy5rrmlleoo3a

ESPNetv2: A Light-weight, Power Efficient, and General Purpose Convolutional Neural Network [article]

Sachin Mehta, Mohammad Rastegari, Linda Shapiro, Hannaneh Hajishirzi
2019 arXiv   pre-print
Our network uses group point-wise and depth-wise dilated separable convolutions to learn representations from a large effective receptive field with fewer FLOPs and parameters.  ...  Experiments on these tasks, including image classification on the ImageNet and language modeling on the PenTree bank dataset, demonstrate the superior performance of our method over the state-of-the-art  ...  Disclaimer: The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing endorsements, either expressed or implied, of IARPA, DOI/IBC, or  ... 
arXiv:1811.11431v3 fatcat:ylym4agztzfxznnf2sxs5qlrce

Quantum Dilated Convolutional Neural Networks

Yixiong Chen
2022 IEEE Access  
Our method extends the concept of dilated convolution, which has been widely applied in modern deep learning algorithms, to the context of hybrid neural networks.  ...  We perform empirical experiments on MNIST and Fashion-MNIST datasets for the task of image recognition and demonstrate that QDCNN models generally enjoy better performances in terms of both accuracy and  ...  It is noteworthy that even though the quantum dilated convolution is able to expand the receptive field the number of data points that are fed into the quantum convolution circuit is the same as the one  ... 
doi:10.1109/access.2022.3152213 fatcat:y7vupsmyandb7hnm5u4qxd7dze

C3: Concentrated-Comprehensive Convolution and its application to semantic segmentation [article]

Hyojin Park, Youngjoon Yoo, Geonseok Seo, Dongyoon Han, Sangdoo Yun, Nojun Kwak
2019 arXiv   pre-print
The second stage increases the receptive field by using a depth-wise separable dilated convolution from the feature map of the first stage.  ...  One of the practical choices for making a lightweight semantic segmentation model is to combine a depth-wise separable convolution with a dilated convolution.  ...  The comprehensive convolution stage uses dilation to see the large receptive field, followed by the point-wise convolution that mixes the channel information.  ... 
arXiv:1812.04920v3 fatcat:ostkgdbz45e7hjdzpw7qh67hrq

Understanding Convolution for Semantic Segmentation [article]

Panqu Wang, Pengfei Chen, Ye Yuan, Ding Liu, Zehua Huang, Xiaodi Hou, Garrison Cottrell
2018 arXiv   pre-print
This framework 1) effectively enlarges the receptive fields (RF) of the network to aggregate global information; 2) alleviates what we call the "gridding issue" caused by the standard dilated convolution  ...  We evaluate our approaches thoroughly on the Cityscapes dataset, and achieve a state-of-art result of 80.1% mIOU in the test set at the time of submission.  ...  Acknowledgments We thank the members of TuSimple and Gary's Unbelievable Research Unit (GURU) for comments on this work.  ... 
arXiv:1702.08502v3 fatcat:cgmtdmjhozechgh3qipjr73egy

Dilated Skip Convolution for Facial Landmark Detection

Seyha Chim, Jin-Gu Lee, Ho-Hyun Park
2019 Sensors  
convolution filters and (2) a dilated skip convolution subnet—a combination of dilated convolutions and skip-connections networks—that are in charge of robustly refining the local appearance heatmaps.  ...  Through this proposed architecture, we demonstrate that our approach achieves state-of-the-art performance on challenging datasets—including LFPW, HELEN, 300W and AFLW2000-3D—by leveraging fully convolutional  ...  Figure 2 illustrates how dilated convolutions enlarge the receptive fields by altering dilation factors (d).  ... 
doi:10.3390/s19245350 pmid:31817213 pmcid:PMC6960628 fatcat:gxpwzfeblvgylnazw3ra6vu5iu

Submanifold Sparse Convolutional Networks [article]

Benjamin Graham, Laurens van der Maaten
2017 arXiv   pre-print
"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 with state-of-the-art methods whilst requiring substantially less computation.  ...  The difference between both operations is in how they handle active sites: instead of automatically making a site active if any of the inputs to its receptive field is active (thereby dilating the set  ... 
arXiv:1706.01307v1 fatcat:qiiqarroe5gujmxnrhn5vi5hyu
« Previous Showing results 1 — 15 out of 7,125 results