Filters








39,691 Hits in 6.3 sec

Convolutional Neural Network Based on Extreme Learning Machine for Maritime Ships Recognition in Infrared Images

Atmane Khellal, Hongbin Ma, Qing Fei
2018 Sensors  
To overcome these weaknesses, we introduce a new approach fully based on Extreme Learning Machine (ELM) to learn useful CNN features and perform a fast and accurate classification, which is suitable for  ...  The proposed approach combines an ELM based learning algorithm to train CNN for discriminative features extraction and an ELM based ensemble for classification.  ...  The ELM-CNN Approach We first start with the ELM-CNN algorithm, which is a new framework based on extreme learning machines able to train any convolution neural network.  ... 
doi:10.3390/s18051490 pmid:29747439 pmcid:PMC5982679 fatcat:j22dpcc6gvafpghsj2a4y4dqem

Light-Head R-CNN: In Defense of Two-Stage Object Detector [article]

Zeming Li, Chao Peng, Gang Yu, Xiangyu Zhang, Yangdong Deng, Jian Sun
2017 arXiv   pre-print
We propose a new two-stage detector, Light-Head R-CNN, to address the shortcoming in current two-stage approaches.  ...  Faster R-CNN involves two fully connected layers for RoI recognition, while R-FCN produces a large score maps.  ...  Having these issues in mind, in our new Light-Head R-CNN, we propose to utilize a simple, cheap fully-connected layer for our R-CNN subnet, which makes a good trade-off between the performance and computational  ... 
arXiv:1711.07264v2 fatcat:tnhsjyytiraapdkmghdsbiktsi

Neural Networks for Parameter Estimation in Intractable Models [article]

Amanda Lenzi, Julie Bessac, Johann Rudi, Michael L. Stein
2021 arXiv   pre-print
Our neural-network-based method provides a competitive alternative to current approaches, as demonstrated by considerable accuracy and computational time improvements.  ...  It serves as a proof of concept for deep learning in statistical parameter estimation and can be extended to other estimation problems.  ...  For instance, consider models for multivariate extremes, where data are usually sampled at a large number of spatial locations but large computational complexities limit their usefulness.  ... 
arXiv:2107.14346v1 fatcat:3ksmufxckrdyriqojcnpcih4qy

Guest Editorial Deep Learning in Medical Imaging: Overview and Future Promise of an Exciting New Technique

Hayit Greenspan, Bram van Ginneken, Ronald M. Summers
2016 IEEE Transactions on Medical Imaging  
Second, training a deep CNN requires large computational and memory resources, without which the training process would be extremely time-consuming.  ...  for a new medical task at hand.  ... 
doi:10.1109/tmi.2016.2553401 fatcat:butvgal6rzethin4oujxzax3wu

Predicting invasive ductal carcinoma using a Reinforcement Sample Learning Strategy using Deep Learning [article]

Rushabh Patel
2021 arXiv   pre-print
We are proposing a method for Invasive ductal carcinoma that will use convolutional neural networks (CNN) on mammograms to assist radiologists in diagnosing the disease.  ...  The aim of this study is to establish a novel method for fully automated feature extraction and classification in invasive ductal carcinoma computer-aided diagnosis (CAD) systems.  ...  To begin, a new CNN model is developed to automatically extract and classify features from mammograms with large image sizes.  ... 
arXiv:2105.12564v2 fatcat:sqzaef5u2radflznjkel26adde

Texture Classification in Extreme Scale Variations using GANet [article]

Li Liu, Jie Chen, Guoying Zhao, Paul Fieguth, Xilin Chen, Matti Pietikäinen
2018 arXiv   pre-print
Because extreme scale variations are not necessarily present in most standard texture databases, to support the proposed extreme-scale aspects of texture understanding we are developing a new dataset,  ...  Motivated by the challenges posed by this problem, we propose a new GANet network where we use a Genetic Algorithm to change the units in the hidden layers during network training, in order to promote  ...  Thus, we have a new training/testing set Θ + I = {I}. 3) For each image in the new training/testing set, we use the proposed FV-GANet to extract global texture feature representation, following FV-CNN  ... 
arXiv:1802.04441v1 fatcat:nexlgfpwxvfbtfxs3ymhcl2svq

Improved deep learning-based macromolecules structure classification from electron cryo-tomograms

Chengqian Che, Ruogu Lin, Xiangrui Zeng, Karim Elmaaroufi, John Galeotti, Min Xu
2018 Machine Vision and Applications  
Specifically, we proposed a deep learning-based image classification approach for large-scale systematic macromolecular structure separation from CECT data.  ...  With recent breakthroughs in Cellular Electron Cryo-Tomography (CECT) 3D imaging technology, it is now possible for researchers to gain accesses to fully study and understand the macro-molecular structures  ...  From that on, CNNs have become a household name in computer vision computer community.  ... 
doi:10.1007/s00138-018-0949-4 pmid:31511756 pmcid:PMC6738941 fatcat:qh5kxzmyerhwnp3fj63lttbdki

FaceDetectNet: face detection via fully-convolutional network

V.S. Gorbatsevich, A.S. Moiseenko, Y.V. Vizilter
2019 Computer Optics  
There are a lot of face detection approaches proposed including different CNN-based techniques, but the problem of optimal balancing between detection quality and computational speed is still relevant.  ...  In this paper we propose new CNN-based solution for face detection called FaceDetectNet.  ...  Different modifications of this approach [7] are still of use until these days mainly due to their extremely high computational speed.  ... 
doi:10.18287/2412-6179-2019-43-1-63-71 fatcat:o3grke5hhffg7iqbhbqo5jjif4

Text-Attentional Convolutional Neural Network for Scene Text Detection

Tong He, Weilin Huang, Yu Qiao, Jian Yao
2016 IEEE Transactions on Image Processing  
In this work, we present a new system for scene text detection by proposing a novel Text-Attentional Convolutional Neural Network (Text-CNN) that particularly focuses on extracting text-related regions  ...  We develop a new learning mechanism to train the Text-CNN with multi-level and rich supervised information, including text region mask, character label, and binary text/nontext information.  ...  Applying a deep CNN for text/non-text classification is not new, but to our knowledge, this is the first attempt to design a CNN model specially for text-related region and text feature computing, by leveraging  ... 
doi:10.1109/tip.2016.2547588 pmid:27093723 fatcat:bdtxestldvf5hlgb5ecx7srtyy

Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning [article]

Michael S. Ryoo, Kiyoon Kim, Hyun Jong Yang
2018 arXiv   pre-print
This paper presents an approach for recognizing human activities from extreme low resolution (e.g., 16x12) videos.  ...  We design a new two-stream multi-Siamese convolutional neural network.  ...  Acknowledgement This research was conducted as a part of EgoVid Inc.'s research activity on privacy-preserving computer vision.  ... 
arXiv:1708.00999v2 fatcat:q55axmk2hfhhbiq2mu3erf7pwi

Reading Scene Text in Deep Convolutional Sequences [article]

Pan He, Weilin Huang, Yu Qiao, Chen Change Loy, Xiaoou Tang
2015 arXiv   pre-print
Then a deep recurrent model, building on long short-term memory (LSTM), is developed to robustly recognize the generated CNN sequences, departing from most existing approaches recognising each character  ...  it to work reliably without either pre- or post-processing; (ii) the deep CNN feature is robust to various image distortions; (iii) it retains the explicit order information in word image, which is essential  ...  They explored the CNN for computing a deep feature from a whole image, followed by a RNN to decode it into a sequence of words.  ... 
arXiv:1506.04395v2 fatcat:zwa65kok6rbzjpyxiwlhqokx2i

Reading Scene Text in Deep Convolutional Sequences

Pan He, Weilin Huang, Yu Qiao, Chen Loy, Xiaoou Tang
2016 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Then a deep recurrent model, building on long short-term memory (LSTM), is developed to robustly recognize the generated CNN sequences, departing from most existing approaches recognising each character  ...  it to work reliably without either pre- or post-processing; (ii) the deep CNN feature is robust to various image distortions; (iii) it retains the explicit order information in word image, which is essential  ...  They explored the CNN for computing a deep feature from a whole image, followed by a RNN to decode it into a sequence of words.  ... 
doi:10.1609/aaai.v30i1.10465 fatcat:itw3iqu2o5fkvestpldanqqpqe

Efficient Correction for EM Connectomics with Skeletal Representation

Konstantin Dimitriev, Toufiq Parag, Brian Matejek, Arie E. Kaufman, Hanspeter Pfister
2018 British Machine Vision Conference  
A reduction of the search space by several orders of magnitude enables our approach to be scalable for terabyte or petabyte scale neuron reconstruction.  ...  In this paper, we present an efficient correction algorithm for EM neuron reconstruction. Each region in a 3D segmentation is represented by its skeleton.  ...  Additional support was provided by the Center for Biotechnology, a New York State Center for Advanced Technology; Cold Spring Harbor Laboratory; Brookhaven National Laboratory; the Feinstein Institute  ... 
dblp:conf/bmvc/DimitrievPMKP18 fatcat:53l77jjbmjdkvo2srv3hxq5kd4

Fusion and binarization of CNN features for robust topological localization across seasons

Roberto Arroyo, Pablo F. Alcantarilla, Luis M. Bergasa, Eduardo Romera
2016 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)  
In this paper, we present a new proposal focused on automatically learned descriptors, which are processed by means of a technique recently popularized in the computer vision community: Convolutional Neural  ...  In addition, we compress the redundant data of CNN features into a tractable number of bits for efficient and robust place recognition.  ...  On the other hand, the approach defined in [27] performs end-to-end learning of a CNN for identifying places.  ... 
doi:10.1109/iros.2016.7759685 dblp:conf/iros/ArroyoABR16 fatcat:o7zjlgluazepvih5tsfwaqrfqi

HPC AI500: A Benchmark Suite for HPC AI Systems [article]

Zihan Jiang, Wanling Gao, Lei Wang, Xingwang Xiong, Yuchen Zhang, Xu Wen, Chunjie Luo, Hainan Ye, Yunquan Zhang, Shengzhong Feng, Kenli Li, Weijia Xu (+1 others)
2019 arXiv   pre-print
The community needs a new yard stick for evaluating the future HPC systems. In this paper, we propose HPC AI500 — a benchmark suite for evaluating HPC systems that running scientific DL workloads.  ...  We propose a set of metrics for comprehensively evaluating the HPC AI systems, considering both accuracy, performance as well as power and cost.  ...  In fact, accurately classifying these jet-images is the key to find signals of new particles.  ... 
arXiv:1908.02607v3 fatcat:bv4aia7yrbglrjtg47zetxmsnu
« Previous Showing results 1 — 15 out of 39,691 results