460 Hits in 5.1 sec

An Energy-Efficient Edge Computing Paradigm for Convolution-based Image Upsampling [article]

Ian Colbert, Ken Kreutz-Delgado, Srinjoy Das
2021 arXiv   pre-print
A novel energy-efficient edge computing paradigm is proposed for real-time deep learning-based image upsampling applications.  ...  the edge improves both system latency and energy efficiency when compared to their sub-pixel or resize convolution counterparts.  ...  We propose a novel edge computing paradigm for real-time convolution-based image upsampling applications that separately considers algorithms for training in the cloud and inference at the edge.  ... 
arXiv:2107.07647v2 fatcat:pz2iab3opbbcthxtbtq4orcnre

Depth Map Super-Resolution via Cascaded Transformers Guidance

Ido Ariav, Israel Cohen
2022 Frontiers in Signal Processing  
Several methods have recently been proposed for guided super-resolution of depth maps using convolutional neural networks to overcome this limitation.  ...  A cascaded transformer module incorporates high-resolution structural information from the intensity image into the depth upsampling process.  ...  The color features are then passed through an edge attention mechanism to highlight the edges useful for upsampling.  ... 
doi:10.3389/frsip.2022.847890 fatcat:bzrqbsx72fgcvoir267666onpe

Deep Affinity Net: Instance Segmentation via Affinity [article]

Xingqian Xu, Mang Tik Chiu, Thomas S. Huang, Honghui Shi
2020 arXiv   pre-print
Despite the maturity of these two paradigms, we would like to report an alternative affinity-based paradigm where instances are segmented based on densely predicted affinities and graph partitioning algorithms  ...  In this work, we propose Deep Affinity Net, an effective affinity-based approach accompanied with a new graph partitioning algorithm Cascade-GAEC.  ...  () > threshold do Our approach We proposed a novel grouping mechanism for our affinity-based instance segmentation network, which is followed by an efficient greedy-based Cascade-GAEC algorithm for graph  ... 
arXiv:2003.06849v1 fatcat:7u2yidtuwbcdjidvkt7p2knjoq

Rate-energy-accuracy optimization of convolutional architectures for face recognition

L. Bondi, L. Baroffio, M. Cesana, M. Tagliasacchi, G. Chiachia, A. Rocha
2016 Journal of Visual Communication and Image Representation  
Face recognition systems based on Convolutional Neural Networks (CNNs) or Convolutional architectures currently represent the state of the art, achieving an accuracy comparable to that of humans.  ...  Therefore, in this paper we address the problem of optimizing the energy-rate-accuracy characteristics of a convolutional architecture for face recognition.  ...  As a first step towards an energy-accuracy optimized model, we disabled the initial image upsampling process and we removed the last layer of the network, that accounts to more than 20% of the total computational  ... 
doi:10.1016/j.jvcir.2015.12.015 fatcat:t7kgz7ui5vdlxoruwz7imntwhu

An Approximate GEMM Unit for Energy-Efficient Object Detection

Ratko Pilipović, Vladimir Risojević, Janko Božič, Patricio Bulić, Uroš Lotrič
2021 Sensors  
Edge computing brings artificial intelligence algorithms and graphics processing units closer to data sources, making autonomy and energy-efficient processing vital for their design.  ...  Approximate computing has emerged as a popular strategy for energy-efficient circuit design, where the challenge is to achieve the best tradeoff between design efficiency and accuracy.  ...  Approximate computing is a new paradigm where an acceptable error is induced in the computing to achieve more energy-efficient processing [28] [29] [30] [31] [32] [33] .  ... 
doi:10.3390/s21124195 fatcat:6wpmpcan5bhlli6zdgrlrg7hpq

Fast Global Image Smoothing Based on Weighted Least Squares

Dongbo Min, Sunghwan Choi, Jiangbo Lu, Bumsub Ham, Kwanghoon Sohn, Minh N. Do
2014 IEEE Transactions on Image Processing  
This paper presents an efficient technique for performing a spatially inhomogeneous edge-preserving image smoothing, called fast global smoother.  ...  Our approach combines the best of two paradigms, i.e., efficient edge-preserving filters and optimization-based smoothing.  ...  CONCLUSIONS This paper has presented an efficient edge-preserving smoothing method based on the WLS formulation, called fast global smoother.  ... 
doi:10.1109/tip.2014.2366600 pmid:25373085 fatcat:e7fngjxyl5bejeahwwlusxttbi

Multimedia super-resolution via deep learning: A survey

Khizar Hayat
2018 Digital signal processing (Print)  
The recent phenomenal interest in convolutional neural networks (CNNs) must have made it inevitable for the super-resolution (SR) community to explore its potential.  ...  We focus on the three important aspects of multimedia - namely image, video and multi-dimensions, especially depth maps.  ...  The architecture (Fig. 12) comprises of an input ReLU based convolution layer, followed by a series of alternating upsampling and convolutional layers, and an output layer.  ... 
doi:10.1016/j.dsp.2018.07.005 fatcat:bhzritty4fcvhay4v2iptj5kge

Deep Learning on Edge TPUs [article]

Andreas M Kist
2021 arXiv   pre-print
The Google Edge TPU is an emerging hardware accelerator that is cost, power and speed efficient, and is available for prototyping and production purposes.  ...  Computing at the edge is important in remote settings, however, conventional hardware is not optimized for utilizing deep neural networks.  ...  Acknowledgment AMK thanks Michael Döllinger, Tobias Schraut and René Groh for their critical comments on the manuscript.  ... 
arXiv:2108.13732v1 fatcat:jfywtr7dpfcmbphdaah7atzpkq

Graph Generative Models for Fast Detector Simulations in High Energy Physics [article]

Ali Hariri, Darya Dyachkova, Sergei Gleyzer
2021 arXiv   pre-print
Accurate and fast simulation of particle physics processes is crucial for the high-energy physics community.  ...  We discuss a graph generative model that provides effective reconstruction of LHC events, paving the way for full detector level fast simulation for HL-LHC.  ...  By comparison, spatial convolution architectures are more efficient in time and memory complexity than spectral-based convolution as shown in [47] .  ... 
arXiv:2104.01725v2 fatcat:mkuiqjvsezhnhgh7jjbhd62iii

RGPNet: A Real-Time General Purpose Semantic Segmentation [article]

Elahe Arani, Shabbir Marzban, Andrei Pata, Bahram Zonooz
2020 arXiv   pre-print
RGPNet consists of a light-weight asymmetric encoder-decoder and an adaptor.  ...  Moreover, towards green AI, we show that using an optimized label-relaxation technique with progressive resizing can reduce the training time by up to 60% while preserving the performance.  ...  We combat aliasing effect in label map on lower resolutions by employing a modified label relaxation • We optimize RGPNet for deployment on an edge computing device using TensorRT, a platform for highperformance  ... 
arXiv:1912.01394v2 fatcat:ycxneyuqk5ejrjo3jg4tfbag44

Image and Video Compression with Neural Networks: A Review

Siwei Ma, Xinfeng Zhang, Chuanmin Jia, Zhenghui Zhao, Shiqi Wang, Shanshe Wanga
2019 IEEE transactions on circuits and systems for video technology (Print)  
The evolution and development of neural network based compression methodologies are introduced for images and video respectively.  ...  novel and promising solution for image and video compression.  ...  In [104] , they provided an efficient solution for CNN based loop filters with memory efficiency.  ... 
doi:10.1109/tcsvt.2019.2910119 fatcat:ibwmmewdlfcexjxfetsxzga52y

StereoSpike: Depth Learning with a Spiking Neural Network [article]

Ulysse Rançon, Javier Cuadrado-Anibarro, Benoit R. Cottereau, Timothée Masquelier
2021 arXiv   pre-print
Depth estimation is an important computer vision task, useful in particular for navigation in autonomous vehicles, or for object manipulation in robotics.  ...  This means that StereoSpike could be efficiently implemented on neuromorphic chips, opening the door for low power and real time embedded systems.  ...  We would like to thank Amirreza Yousefzadeh for his help and expertise on digital neuromorphic hardware.  ... 
arXiv:2109.13751v2 fatcat:hcghwoxegzgeljc5oj7zb6sa5m

Saliency Unified: A Deep Architecture for simultaneous Eye Fixation Prediction and Salient Object Segmentation

Srinivas S. S. Kruthiventi, Vennela Gudisa, Jaley H. Dholakiya, R. Venkatesh Babu
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
In addition, our network captures saliency at multiple scales via inceptionstyle convolution blocks.  ...  Our network shows a significant improvement over the current state-of-the-art for both eye fixation prediction and salient object segmentation across a number of challenging datasets.  ...  We thank Nvidia for their hardware grant and Google for the travel grant.  ... 
doi:10.1109/cvpr.2016.623 dblp:conf/cvpr/KruthiventiGDB16 fatcat:5qdfu53dgrfxvg2ly3n6xgxhte

Recent Advances in the Applications of Convolutional Neural Networks to Medical Image Contour Detection [article]

Zizhao Zhang and Fuyong Xing and Hai Su and Xiaoshuang Shi and Lin Yang
2017 arXiv   pre-print
Deep convolution neural networks (CNNs), as one of the most important branch of the deep learning family, have been widely investigated for various computer-aided diagnosis tasks including long-term problems  ...  Image contour detection is a fundamental but challenging task that has been studied for more than four decades.  ...  Deformable model based methods focus on deforming an initial active contour to align the object boundary by solving an energy function.  ... 
arXiv:1708.07281v1 fatcat:kdplgrjf4zaurcdbszcoeinktm

Continuous Conditional Random Field Convolution for Point Cloud Segmentation

Fei Yang, Franck Davoine, Huan Wang, Zhong Jin
2021 Pattern Recognition  
Furthermore, we build an encoder-decoder network based on the proposed continuous CRF graph convolution (CRFConv), in which the CRFConv embedded in the decoding layers can restore the details of high-level  ...  Therefore, we first model the point cloud features with a continuous quadratic energy model and formulate its solution process as a message-passing graph convolution, by which it can be easily integrated  ...  Natural Science Foundation of China under Grant Nos 61872188, 61703209, U1713208, 61972204, 61672287, 61861136011, 61773215, and by the French Labex MS2T ANR-11-IDEX-0004-02 through the program Investments for  ... 
doi:10.1016/j.patcog.2021.108357 fatcat:eyz6xdy5dzf3npmcflx4zdhvtu
« Previous Showing results 1 — 15 out of 460 results