2,319 Hits in 4.9 sec

Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs [article]

Liang-Chieh Chen and George Papandreou and Iasonas Kokkinos and Kevin Murphy and Alan L. Yuille
2016 arXiv   pre-print
We overcome this poor localization property of deep networks by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF).  ...  Deep Convolutional Neural Networks (DCNNs) have recently shown state of the art performance in high level vision tasks, such as image classification and object detection.  ...  with the fully-connected CRF.  ... 
arXiv:1412.7062v4 fatcat:qntjxk62unge5o5bm4vqbctism

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs [article]

Liang-Chieh Chen and George Papandreou and Iasonas Kokkinos and Kevin Murphy and Alan L. Yuille
2017 arXiv   pre-print
In this work we address the task of semantic image segmentation with Deep Learning and make three main contributions that are experimentally shown to have substantial practical merit.  ...  We overcome this by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF), which is shown both qualitatively and quantitatively to improve localization performance  ...  To produce semantically accurate predictions and detailed segmentation maps along object boundaries, we also combine ideas from deep convolutional neural networks and fully-connected conditional random  ... 
arXiv:1606.00915v2 fatcat:x76tf4nbsfhhhnplubxdtjl6qy

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

Liang-Chieh Chen, George Papandreou, Iasonas Kokkinos, Kevin Murphy, Alan L. Yuille
2018 IEEE Transactions on Pattern Analysis and Machine Intelligence  
We incorporate such masks in CNNs and replace the convolution operation with a "segmentationaware" variant that allows a neuron to selectively attend to inputs coming from its own region.  ...  Our results show that in semantic segmentation we can match the performance of DenseCRFs while being faster and simpler, and in optical flow we obtain clearly sharper responses than networks that do not  ...  This is a fully-convolutional network, with a contractive part that reduces the resolution of the input by a factor of 64, and an expansionary part (with skip connections) that restores the resolution  ... 
doi:10.1109/tpami.2017.2699184 pmid:28463186 fatcat:44cino7l4jfjngddvqw44srupm

W-Net: A Deep Model for Fully Unsupervised Image Segmentation [article]

Xide Xia, Brian Kulis
2017 arXiv   pre-print
We borrow recent ideas from supervised semantic segmentation methods, in particular by concatenating two fully convolutional networks together into an autoencoder--one for encoding and one for decoding  ...  In this paper, we revisit the problem of purely unsupervised image segmentation and propose a novel deep architecture for this problem.  ...  Further, Figure [11] and Figure [12] illustrates more results of running the W-Net+ucm on images from the BSDS500.  ... 
arXiv:1711.08506v1 fatcat:ev74kdmwynhgpdkk7z4u2vr62a

Gaussian Filter in CRF Based Semantic Segmentation [article]

Yichi Gu, Qisheng Wu, Jing Li, Kai Cheng
2017 arXiv   pre-print
Fully convolutional network [1] is the standard model for semantic segmentation.  ...  In this paper, we introduce a multi-resolution neural network for FCN and apply Gaussian filter to the extended CRF kernel neighborhood and the label image to reduce the oscillating effect of CRF neural  ...  Combination of FCN and CRF The semantic segmentation network FCN is connected with CRF by fully connection.  ... 
arXiv:1709.00516v1 fatcat:yjfivat6yjbelknyx5neb3e56a

Exploring Context with Deep Structured models for Semantic Segmentation [article]

Guosheng Lin, Chunhua Shen, Anton van den Hengel, Ian Reid
2017 arXiv   pre-print
State-of-the-art semantic image segmentation methods are mostly based on training deep convolutional neural networks (CNNs).  ...  In this work, we proffer to improve semantic segmentation with the use of contextual information. In particular, we explore 'patch-patch' context and 'patch-background' context in deep CNNs.  ...  method is a Potts-model-based fully-connected CRF with pairwise potentials based on color contrast for local smoothness.  ... 
arXiv:1603.03183v3 fatcat:tmfows5y5velxitkap7oso2kn4

Edge Prior Multilayer Segmentation Network Based on Bayesian Framework

Chu He, Zishan Shi, Peizhang Fang, Dehui Xiong, Bokun He, Mingsheng Liao
2020 Journal of Sensors  
This paper proposes an edge prior semantic segmentation architecture based on Bayesian framework.  ...  In recent years, methods based on neural network have achieved excellent performance for image segmentation.  ...  41371342 and 61331016), and the Hubei Innovation Group (grant number 2018CFA006).  ... 
doi:10.1155/2020/6854260 fatcat:6jg2fi73cvce7hy5a4xtxccm2m

Squeeze-SegNet: A new fast Deep Convolutional Neural Network for Semantic Segmentation [article]

Geraldin Nanfack, Azeddine Elhassouny, Rachid Oulad Haj Thami
2017 arXiv   pre-print
We present a new Deep fully Convolutional Neural Network for pixel-wise semantic segmentation which we call Squeeze-SegNet. The architecture is based on Encoder-Decoder style.  ...  object detection and pixel-wise semantic segmentation.  ...  Thanking the Lord God, We would also like to thank the coordination of the Research Master in Data Sciences and Big Data of ENSIAS and then in another measure the family and the future Doctor Harry Kamdem  ... 
arXiv:1711.05491v1 fatcat:tmoih4ptsndytgij5fkrg3bipi

A Survey on Deep Learning-based Architectures for Semantic Segmentation on 2D images [article]

Irem Ulku, Erdem Akagunduz
2022 arXiv   pre-print
Boosted by the extraordinary ability of convolutional neural networks (CNN) in creating semantic, high level and hierarchical image features; several deep learning-based 2D semantic segmentation approaches  ...  We started with an analysis of the public image sets and leaderboards for 2D semantic segmentation, with an overview of the techniques employed in performance evaluation.  ...  (w/o course) 39.3% mIoU @ADE20K DeepLab.v1 CNN with dilated convolutions, succeeded by a fully-connected (i.e. Dense) CRF.  ... 
arXiv:1912.10230v4 fatcat:3y74apt2hnaypcdkmjs3sf4zb4

Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical Imagery

Daliana Lobo Torres, Raul Queiroz Feitosa, Patrick Nigri Happ, Laura Elena Cué La Rosa, José Marcato Junior, José Martins, Patrik Olã Bressan, Wesley Nunes Gonçalves, Veraldo Liesenberg
2020 Sensors  
This study proposes and evaluates five deep fully convolutional networks (FCNs) for the semantic segmentation of a single tree species: SegNet, U-Net, FC-DenseNet, and two DeepLabv3+ variants.  ...  We also verify the benefits of fully connected conditional random fields (CRFs) as a post-processing step to improve the segmentation maps.  ...  deep learning semantic segmentation methods, namely U-Net, SegNet, FC-DenseNet, and Deeplabv3+ with the Xception and MobileNetV2 backbone, for the segmentation of cumbaru trees on the aforementioned RGB  ... 
doi:10.3390/s20020563 pmid:31968589 pmcid:PMC7014541 fatcat:gsrr45wpvrhojm2k3wksinxmpy

Building Segmentation through a Gated Graph Convolutional Neural Network with Deep Structured Feature Embedding [article]

Yilei Shi, Qingyu Li, Xiao Xiang Zhu
2019 arXiv   pre-print
Semantic segmentation is an efficient approach for this task. The latest development in deep convolutional neural networks (DCNNs) has made accurate pixel-level classification tasks possible.  ...  Deep architectures generally fail to produce fine-grained segmentation with accurate boundaries due to their progressive down-sampling.  ...  Murphy and A. L. Yuille, “DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs,” IEEE Trans. Pattern Anal. Mach.  ... 
arXiv:1911.03165v1 fatcat:jmg2fbqomzht7cgnzuqhtz2sw4

Face Parsing via a Fully-Convolutional Continuous CRF Neural Network [article]

Lei Zhou, Zhi Liu, Xiangjian He
2017 arXiv   pre-print
In this work, we address the face parsing task with a Fully-Convolutional continuous CRF Neural Network (FC-CNN) architecture.  ...  In contrast to previous face parsing methods that apply region-based subnetwork hundreds of times, our FC-CNN is fully convolutional with high segmentation accuracy.  ...  We compare the proposed FC-CNN with a number of recent fully convolutional semantic segmentation methods with competitive performance.  ... 
arXiv:1708.03736v1 fatcat:mwrcdmwtijcnzmc4mqswzvby2m

DNS: A multi-scale deconvolution semantic segmentation network for joint detection and segmentation

Ning Feng, Le Dong, Qianni Zhang, Ning Zhang, Xi Wu, Jianwen Chen, W. Anggono
2019 MATEC Web of Conferences  
Real-time semantic segmentation has become crucial in many applications such as medical image analysis and autonomous driving.  ...  In this paper, we introduce a single semantic segmentation network, called DNS, for joint object detection and segmentation task.  ...  Although post-processing the output of FCN with a fully-connected CRF can increase segmentation accuracy near object boundaries, mean-field inference in fully-connected CRF model is expensive in terms  ... 
doi:10.1051/matecconf/201927702005 fatcat:y4oyemcmhfb3pj7hoko4yybe5a

Learning deep representations for semantic image parsing: a comprehensive overview

Lili Huang, Jiefeng Peng, Ruimao Zhang, Guanbin Li, Liang Lin
2018 Frontiers of Computer Science  
In this paper, we summarize three aspects of the progress of research on semantic image parsing, i.e., category-level semantic segmentation, instance-level semantic segmentation, and beyond segmentation  ...  Semantic image parsing, which refers to the process of decomposing images into semantic regions and constructing the structure representation of the input, has recently aroused widespread interest in the  ...  Another fundamental work−−DeepLab system [13] −− integrated CNN with fully connected conditional random field (CRF) to expand and improve FCN [12] .  ... 
doi:10.1007/s11704-018-7195-8 fatcat:p5hvfwhl5rbork5vf4rpnx3h6u

Efficient Yet Deep Convolutional Neural Networks for Semantic Segmentation [article]

Sharif Amit Kamran, Ali Shihab Sabbir
2018 arXiv   pre-print
Semantic Segmentation using deep convolutional neural network pose more complex challenge for any GPU intensive task.  ...  With the introduction of Fully Convolutional Neural Network, which uses finer strides and utilizes deconvolutional layers for upsampling, it has been a go to for any image segmentation task.  ...  ACKNOWLEDGMENT We would like to thank Evan Shelhamer for providing the evaluation scripts and Caffe users community for their advice and suggestions.  ... 
arXiv:1707.08254v3 fatcat:6mvr3xumerh5tnw45zakghqm2e
« Previous Showing results 1 — 15 out of 2,319 results