Filters








29,719 Hits in 4.0 sec

Fully Convolutional Network and Region Proposal for Instance Identification with Egocentric Vision

Maxime Portaz, Matthias Kohl, Georges Quenot, Jean-Pierre Chevallet
2017 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)  
This approach uses fully convolutional networks (FCN) to obtain region proposals without the need for an additional component in the network and training.  ...  It is particularly suited for small datasets with low object variability. The proposed network can be trained end-to-end and produces an effective global descriptor as an image representation.  ...  Fully Convolutional Network Thus, the idea is to start by fine-tuning a network with images at different scales. This can be achieved by using a fully convolutional network (FCN) [17] .  ... 
doi:10.1109/iccvw.2017.281 dblp:conf/iccvw/PortazKQC17 fatcat:33o4i5qiujfgtke37jgr7bam3u

Literature Review: Human Segmentation with Static Camera [article]

Jiaxin Xu, Rui Wang, Vaibhav Rakheja
2019 arXiv   pre-print
Our research topic is Human segmentation with static camera. This topic can be divided into three sub-tasks, which are object detection, instance identification and segmentation.  ...  In this literature review, we will first introduce the background of human segmentation and then talk about issues related to the above three fields as well as how they interact with each other.  ...  In 2016, Long et al. proposed Fully Convolutional Network(FCN) for Semantic Segmentation [22] .  ... 
arXiv:1910.12945v1 fatcat:cmo3qp4vdbadlkvl4tcxtzy7te

Facing Erosion Identification in Railway Lines Using Pixel-Wise Deep-Based Approaches

Keiller Nogueira, Gabriel L. S. Machado, Pedro H. T. Gama, Caio C. V. da Silva, Remis Balaniuk, Jefersson A. dos Santos
2020 Remote Sensing  
A crucial step for automatic erosion identification is to create a good feature representation. Towards such objective, deep learning can learn data-driven features and classifiers.  ...  Soil erosion is considered one of the most expensive natural hazards with a high impact on several infrastructure assets.  ...  The authors gratefully acknowledge the support of the NVIDIA Corporation with the donation of the GeForce GTX TITAN X GPU to the PATREO Laboratory that were used in this work.  ... 
doi:10.3390/rs12040739 fatcat:7jt73ibufvg67l3npgpap4qhka

Contour Extraction of Individual Cattle from an Image Using Enhanced Mask R-CNN Instance Segmentation Method

Rotimi-Williams Bello, Ahmad Sufril Azlan Mohamed, Abdullah Zawawi Talib
2021 IEEE Access  
Three steps are usually involved in instance segmentation, namely identification of regions of the proposal using region proposal network (RPN), object class prediction, and object mask extraction.  ...  Also, in the study [32] , the authors proposed the extension of a fully convolutional network in livestock practice to achieve beef cattle segmentation.  ... 
doi:10.1109/access.2021.3072636 fatcat:n6nszhd6afhvzh7oxqvfclnuoa

Bird Species Identification Using Yolact Classifier

Sofia K. Pillai
2021 Bioscience Biotechnology Research Communications  
Architectural Overview: The input images are fed into a fully convolutional neural network backbone.  ...  As compared to the fully convolutional instance aware semantic segmentation and region based convolutional neural network, the YOLACT architecture has comparatively less noise and follows a boundary.  ... 
doi:10.21786/bbrc/14.9.23 fatcat:ro46j6welncr5kvaipg775dfci

U-Net Based Multi-instance Video Object Segmentation [article]

Heguang Liu, Jingle Jiang
2019 arXiv   pre-print
In this paper, we implement an effective fully convolutional networks with U-Net similar structure built on top of OSVOS fine-tuned layer.  ...  Multi-instance video object segmentation is to segment specific instances throughout a video sequence in pixel level, given only an annotated first frame.  ...  Also want to thank Google Cloud for sponsoring GPU instances for model training.  ... 
arXiv:1905.07826v1 fatcat:qly6nndj7rh3joajasrupitk6q

AN IMPROVED MULTI-OBJECT INSTANCE SEGMENTATION BASED ON DEEP LEARNING

Nawaf Farhan Funkur Alshdaifat, School of Computer Sciences Universiti Sains Malaysia,11800, Pulau Pinang, Malaysia, Mohd Azam osman, Abdullah Zawawi Talib, School of Computer Sciences Universiti Sains Malaysia,11800, Pulau Pinang, Malaysia, School of Computer Sciences Universiti Sains Malaysia,11800, Pulau Pinang, Malaysia
2021 Maǧallaẗ Al-Kuwayt li-l-ʿulūm  
Firstly, it improves the RestNet-101 (Residual Neural Network) backbone by connecting it to the convolution layer for each ResNet block.  ...  Secondly, the localization of multiple objects is improved by enhancing the Region Proposal Network (RPN), and thirdly, a complex instance segmentation approach is utilized.  ...  The third stage involves adopting the Fully Convolution Network (FCN) to produce an instance segmentation that overcomes the object overlapping problem.  ... 
doi:10.48129/kjs.10879 fatcat:stxg2iy22fbdtb4ekzi3ctocuq

COMPARATIVE ANALYSIS OF DEEP LEARNING ARCHITECTURES FOR GRAPE CLUSTER INSTANCE SEGMENTATION

Ms. Dhanashree Barbole, Dr. Parul Jadhav
2021 Information Technology in Industry  
The grape cluster identification and its segmentation for the sake of total weight prediction task of wine yard shows the need of segmentation atomization with better accuracy.  ...  The challenge of grape cluster segmentation is considered to provide solution using deep neural network models such as YOLO v3, Mask RCNN, U-net.  ...  The deep neural network based grape cluster identification is the instance segmentation task using neural network model.  ... 
doi:10.17762/itii.v9i1.138 fatcat:hkd56qfu45d7zjby5degmcwcia

Shelf Commodity Identification Method Based on Hybrid Fully Convolutional Automatic Encoder

Aofeng Cheng, Guodong Chen, Zheng Wang
2019 IEEE Access  
INDEX TERMS Fully convolutional neural network, autoencoder, object identification, shelf regulation.  ...  In view of the inadequate results of features extracted by these algorithms from RGB image, a hybrid fully convolutional autoencoder neural network (HFCAN) structure, which introduces fully convolutional  ...  Wong et al. proposed a new integrated method for object identification and pose estimation, combining convolutional neural network with multi-hypothesis point cloud registration to achieve robust pixel-based  ... 
doi:10.1109/access.2019.2955560 fatcat:uwshcxv5zjbblfud34jidc7leq

Comparison and Analysis of Algorithms used for Detecting Slums in Satellite Images

Pallavi Saindane, Gayatri Ganapathy, Neha Prabhavalkar, Nilesh Bhatia, Aishwarya Vaidyaa
2019 International Journal of New Technology and Research  
The motivation for this study comes from the potential use of such analysis for identification of slums from satellite images.  ...  Pages 48-52 48 www.ijntr.org  Abstract-This paper presents a comparison of some neural network based approaches for analyzing satellite images.  ...  Fully Convolutional Network Fully Convolutional Networks are used for the purpose of semantic segmentation.  ... 
doi:10.31871/ijntr.5.2.14 fatcat:suxds4ygi5bu5g5lwmurskjmzi

Learning to count with deep object features

Santi Segui, Oriol Pujol, Jordi Vitria
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
Learning to count is a learning strategy that has been recently proposed in the literature for dealing with problems where estimating the number of object instances in a scene is the final objective.  ...  In this paper we explore the features that are learned when training a counting convolutional neural network in order to understand their underlying representation.  ...  We gratefully acknowledge the support of NVIDIA Corporation with the donation of a Tesla K40 GPU used for this research.  ... 
doi:10.1109/cvprw.2015.7301276 dblp:conf/cvpr/SeguiPV15 fatcat:ackgxxegjvfubcxfy4mzwxt5fa

Learning to count with deep object features [article]

Santi Seguí, Oriol Pujol, Jordi Vitrià
2015 arXiv   pre-print
Learning to count is a learning strategy that has been recently proposed in the literature for dealing with problems where estimating the number of object instances in a scene is the final objective.  ...  In this paper we explore the features that are learned when training a counting convolutional neural network in order to understand their underlying representation.  ...  We gratefully acknowledge the support of NVIDIA Corporation with the donation of a Tesla K40 GPU used for this research.  ... 
arXiv:1505.08082v1 fatcat:fmjjbmwbjzezno5ukhc5ynsi7i

Visual Localisation and Individual Identification of Holstein Friesian Cattle via Deep Learning

William Andrew, Colin Greatwood, Tilo Burghardt
2017 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)  
The core architecture consists of 5 stacked convolutional layers, which are shared with the region proposal network -plus two fully connected layers.  ...  Furthermore, Regionbased Fully Convolutional Networks (R-FCNs) [12] -based on fully convolutional architectures, such as FCN [38] -go further and avoid the per-proposal evaluation of Fast and Faster  ... 
doi:10.1109/iccvw.2017.336 dblp:conf/iccvw/AndrewGB17 fatcat:fynpaedybbh2lp7bwzku22b6nu

Intelligent Identification Method of Insulator Defects Based on CenterMask

Zhiming Xuan, Jiwei Ding, Jing Mao
2022 IEEE Access  
Finally, the SAG-Mask with spatial attention mechanism is performed to extract the insulator mask image, while the defect identification and location is realized based on the anchor-free FCOS algorithm  ...  Subsequently, the residual connection and effective Squeeze-Excitation module are introduced to improve the original backbone network, thus overcoming the problem of deep network saturation and channel  ...  It is a fully convolutional one-stage object detection algorithm which realizes the object detection function in the way of per-pixel forecasting.  ... 
doi:10.1109/access.2022.3179975 fatcat:owdgihdiznbu5kgjz4k4wapgj4

Learning to Identify Object Instances by Touch: Tactile Recognition via Multimodal Matching [article]

Justin Lin, Roberto Calandra, Sergey Levine
2019 arXiv   pre-print
The sense of touch can provide robots with an alternative mechanism for recognizing objects. In this paper, we study the problem of touch-based instance recognition.  ...  To our knowledge, our work is the first to address this type of multi-modal instance recognition problem on such a large-scale with our analysis spanning 98 different objects.  ...  ACKNOWLEDGEMENTS We thank Andrew Owens for his insights about multimodal networks and his suggestions for the manuscript.  ... 
arXiv:1903.03591v1 fatcat:2ncg3jdpfbe7bknhh44wbse664
« Previous Showing results 1 — 15 out of 29,719 results