Filters








550 Hits in 4.3 sec

Superpixel Sampling Networks [chapter]

Varun Jampani, Deqing Sun, Ming-Yu Liu, Ming-Hsuan Yang, Jan Kautz
2018 Lecture Notes in Computer Science  
The resulting Superpixel Sampling Network (SSN) is end-to-end trainable, which allows learning task-specific superpixels with flexible loss functions and has fast runtime.  ...  Existing superpixel algorithms are not differentiable, making them difficult to integrate into otherwise end-to-end trainable deep neural networks.  ...  We thank Wei-Chih Tu for providing evaluation scripts. We thank Ben Eckart for his help in the supplementary video.  ... 
doi:10.1007/978-3-030-01234-2_22 fatcat:otvct6i6vbdqnlrdl4ljrn2e5u

Superpixel Sampling Networks [article]

Varun Jampani and Deqing Sun and Ming-Yu Liu and Ming-Hsuan Yang and Jan Kautz
2018 arXiv   pre-print
The resulting "Superpixel Sampling Network" (SSN) is end-to-end trainable, which allows learning task-specific superpixels with flexible loss functions and has fast runtime.  ...  Existing superpixel algorithms are not differentiable, making them difficult to integrate into otherwise end-to-end trainable deep neural networks.  ...  We thank Wei-Chih Tu for providing evaluation scripts. We thank Ben Eckart for his help in the supplementary video. Superpixel Sampling Networks  ... 
arXiv:1807.10174v1 fatcat:phvcryxznbbwpl22sdrplwussm

Generating superpixels using deep image representations [article]

Thomas Verelst, Matthew Blaschko, Maxim Berman
2019 arXiv   pre-print
Many superpixel methods only rely on colors features for segmentation, limiting performance in low-contrast regions and applicability to infrared or medical images where object boundaries have wide appearance  ...  Superpixel algorithms are a common pre-processing step for computer vision algorithms such as segmentation, object tracking and localization.  ...  We research a trainable superpixel algorithm incorporating a neural network that can tune superpixels to a certain image set. A.  ... 
arXiv:1903.04586v1 fatcat:vzdwwmko4jckpoq5xtld5j4kpe

Superpixel Segmentation with Fully Convolutional Networks [article]

Fengting Yang, Qian Sun, Hailin Jin, Zihan Zhou
2020 arXiv   pre-print
Specifically, we modify a popular network architecture for stereo matching to simultaneously predict superpixels and disparities.  ...  In computer vision, superpixels have been widely used as an effective way to reduce the number of image primitives for subsequent processing.  ...  This work is supported in part by NSF award #1815491 and a gift from Adobe.  ... 
arXiv:2003.12929v1 fatcat:muqjv2as3bb3togsae4kowcm4q

Superpixel Segmentation With Fully Convolutional Networks

Fengting Yang, Qian Sun, Hailin Jin, Zihan Zhou
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
Specifically, we modify a popular network architecture for stereo matching to simultaneously predict superpixels and disparities.  ...  In computer vision, superpixels have been widely used as an effective way to reduce the number of image primitives for subsequent processing.  ...  This work is supported in part by NSF award #1815491 and a gift from Adobe.  ... 
doi:10.1109/cvpr42600.2020.01398 dblp:conf/cvpr/YangSJ020 fatcat:2qdi4pplkbgrdmfvrzu3owb7ue

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

Lei Zhou, Zhi Liu, Xiangjian He
2017 arXiv   pre-print
Based on a differentiable superpixel pooling layer and a differentiable C-CRF layer, the unary network and pairwise network are combined via a novel continuous CRF network to achieve spatial consistency  ...  Comprehensive evaluations on LFW-PL and HELEN datasets demonstrate that FC-CNN achieves better performance over the other state-of-arts for accurate face labeling on challenging images.  ...  END-TO-END TRAINING FOR FACE PARSING In this section, we will describe the way for training the network in an end-to-end way.  ... 
arXiv:1708.03736v1 fatcat:mwrcdmwtijcnzmc4mqswzvby2m

Hierarchical Mask Calibration for Unified Domain Adaptive Panoptic Segmentation [article]

Jingyi Zhang, Jiaxing Huang, Shijian Lu
2022 arXiv   pre-print
However, existing studies employ two networks for instance segmentation and semantic segmentation separately which lead to a large amount of network parameters with complicated and computationally intensive  ...  It has three unique features: 1) it enables unified domain adaptive panoptic adaptation; 2) it mitigates false predictions and improves domain adaptive panoptic segmentation effectively; 3) it is end-to-end  ...  from being end-to-end trainable.  ... 
arXiv:2206.15083v1 fatcat:ggx7tvyapzhhvflpz7fv4oqblm

SIN:Superpixel Interpolation Network [article]

Qing Yuan, Songfeng Lu, Yan Huang, Wuxin Sha
2021 arXiv   pre-print
Multi-layer outputs of a fully convolutional network are utilized to predict association scores for interpolations.  ...  In this paper, we propose a deep learning-based superpixel segmentation algorithm SIN which can be integrated with downstream tasks in an end-to-end way.  ...  To form an end-to-end trainable network, SSN [11] turns SLIC into a differentiable algorithm by relaxing the nearest neighbors' constraints.  ... 
arXiv:2110.08702v1 fatcat:g4bbgte6bneedggxqdy4envxkq

Higher Order Conditional Random Fields in Deep Neural Networks [article]

Anurag Arnab, Sadeep Jayasumana, Shuai Zheng, Philip Torr
2016 arXiv   pre-print
In this paper, we demonstrate that two types of higher order potential, based on object detections and superpixels, can be included in a CRF embedded within a deep network.  ...  Recent deep learning approaches have incorporated CRFs into Convolutional Neural Networks (CNNs), with some even training the CRF end-to-end with the rest of the network.  ...  outputs and superpixel segmentation in a CRF which is learned end-to-end in a deep network.  ... 
arXiv:1511.08119v4 fatcat:tku2oa42vrfkjehlrxrkat7wvu

DIC: Deep Image Clustering for Unsupervised Image Segmentation

Lei Zhou, Yufeng Wei
2020 IEEE Access  
The DIC consists of a feature transformation subnetwork (FTS) and a trainable deep clustering subnetwork (DCS) for unsupervised image clustering.  ...  Encouraged by neural networks' flexibility and their ability for modelling intricate patterns, an unsupervised segmentation framework based on a novel deep image clustering (DIC) model is proposed.  ...  the DIC parameters in an end-to-end way, in which the superpixels can serve as the grouping cues for encoding complex image patterns.  ... 
doi:10.1109/access.2020.2974496 fatcat:aburl2wghret7fhbrdw63rzlfm

CRF learning with CNN features for hyperspectral image segmentation

Fahim Irfan Alam, Jun Zhou, Alan Wee-Chung Liew, Xiuping Jia
2016 2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS)  
After a hyperspectral image is over-segmented into superpixels, a deep Convolutional Neural Network (CNN) is used to perform superpixel-level labelling.  ...  This paper proposes a method that uses both spectral and spatial information to segment remote sensing hyperspectral images.  ...  This complete system therefore combines the usefulness of both CNN and CRF and is trainable end-to-end by back propagation algorithm.  ... 
doi:10.1109/igarss.2016.7730798 dblp:conf/igarss/AlamZLJ16 fatcat:ga5vzryxcrch5drbdr27ktseqy

Implicit Integration of Superpixel Segmentation into Fully Convolutional Networks [article]

Teppei Suzuki
2021 arXiv   pre-print
However, to combine superpixels with convolutional neural networks (CNNs) in an end-to-end fashion, one requires extra models to generate superpixels and special operations such as graph convolution.  ...  Superpixels are a useful representation to reduce the complexity of image data.  ...  Unlike existing methods combining superpixel segmentation and neural networks, our method does not use superpixels for downsampling, explicitly.  ... 
arXiv:2103.03435v1 fatcat:t5mzwjbievaipahg7bap5xpjk4

Point Label Aware Superpixels for Multi-species Segmentation of Underwater Imagery [article]

Scarlett Raine, Ross Marchant, Brano Kusy, Frederic Maire, Tobias Fischer
2022 arXiv   pre-print
Our point label aware superpixel method utilizes the sparse point labels, and clusters pixels using learned features to accurately generate single-species segments in cluttered, complex coral images.  ...  Analysis of this imagery can be automated using a model trained to perform semantic segmentation, however it is too costly and time-consuming to densely label images for training supervised models.  ...  an end-to-end trainable, task-specific architecture.  ... 
arXiv:2202.13487v1 fatcat:ehy2dr5ojramrlekvxrpp2vcu4

Learning Hierarchical Features for Scene Labeling

Clement Farabet, Camille Couprie, Laurent Najman, Yann LeCun
2013 IEEE Transactions on Pattern Analysis and Machine Intelligence  
The method alleviates the need for engineered features, and produces a powerful representation that captures texture, shape and contextual information.  ...  Scene labeling consists in labeling each pixel in an image with the category of the object it belongs to.  ...  ACKNOWLEDGMENT We would like to thank Marco Scoffier for fruitful discussions and the 360 degree video collection.  ... 
doi:10.1109/tpami.2012.231 pmid:23787344 fatcat:6juwyqqzw5flxk2ppgsldf6yjm

Semi-Supervised Hierarchical Semantic Object Parsing [article]

Jalal Mirakhorli, Hamidreza Amindavar
2017 arXiv   pre-print
Our model is based on a deep neural network trained for object masking that supervised with input image and follow incorporates a Conditional Random Field (CRF) with end-to-end trainable piecewise order  ...  Models based on Convolutional Neural Networks (CNNs) have been proven very successful for semantic segmentation and object parsing that yield hierarchies of features.  ...  [30] , this can be formulated as a Recurrent Neural Network (RNN), allowing it to be trained end-to-end as part of a larger network.  ... 
arXiv:1709.08019v3 fatcat:tmikuiiftzcs7kirdlhqfuwl4i
« Previous Showing results 1 — 15 out of 550 results