Filters








45,993 Hits in 4.3 sec

Fully Convolutional Multi-Class Multiple Instance Learning [article]

Deepak Pathak, Evan Shelhamer, Jonathan Long, Trevor Darrell
2015 arXiv   pre-print
We propose a novel MIL formulation of multi-class semantic segmentation learning by a fully convolutional network.  ...  Multiple instance learning (MIL) can reduce the need for costly annotation in tasks such as semantic segmentation by weakening the required degree of supervision.  ...  In this work, we propose a novel framework for multiple instance learning (MIL) with a fully convolutional network (FCN).  ... 
arXiv:1412.7144v4 fatcat:2jyzciesorem5ncvnbo6bkjtiu

Real-Time Weakly Supervised Object Detection Using Center-of-Features Localization

Hatem Ibrahem, Ahmed Salem, Hyun Soo Kang
2021 IEEE Access  
INDEX TERMS Object detection, object localization, weakly supervised learning, convolutional neural networks.  ...  The proposed method, called centerof-features localization (COFL), performs localization of objects in a visual scene by combining both multi-label classification and regression for the number of instances  ...  This function learns the multi-label classification task and the class instance count regression task simultaneously rather than separately, as they support each other and help the network learn faster  ... 
doi:10.1109/access.2021.3064372 fatcat:2bvdfr6d7fggjf7qsgmwlrrn4i

Transfer Learning Assisted Classification and Detection of Alzheimer's Disease Stages Using 3D MRI Scans

Maqsood, Nazir, Khan, Aadil, Jamal, Mehmood, Song
2019 Sensors  
The proposed system works on an efficient technique of utilizing transfer learning to classify the images by fine-tuning a pre-trained convolutional network, AlexNet.  ...  The algorithm showed promising results by giving the best overall accuracy of 92.85% for multi-class classification of un-segmented images.  ...  We observe that fully connected layers learn class-specific features to distinguish among classes.  ... 
doi:10.3390/s19112645 pmid:31212698 pmcid:PMC6603745 fatcat:73rdfdcx5nemfhx3iyw4725gz4

Multi-Scale Label Relation Learning for Multi-Label Classification Using 1-Dimensional Convolutional Neural Networks [article]

Junhyung Kim, Byungyoon Park, Charmgil Hong
2021 arXiv   pre-print
We present Multi-Scale Label Dependence Relation Networks (MSDN), a novel approach to multi-label classification (MLC) using 1-dimensional convolution kernels to learn label dependencies at multi-scale  ...  By training a model with multiple kernel sizes, the method is able to learn the dependency relations among labels at multiple scales, while it uses a drastically smaller number of parameters.  ...  Multi-label classification, or MLC in short, is a classification problem where each data instance can be associated with multiple label variables.  ... 
arXiv:2107.05941v1 fatcat:mabcqxdnyvfgjcgg3qebx6qtya

Multi-Temporal Resolution Convolutional Neural Networks for Acoustic Scene Classification [article]

Alexander Schindler, Thomas Lidy, Andreas Rauber
2018 arXiv   pre-print
This architecture is composed of separated parallel Convolutional Neural Networks which learn spectral and temporal representations for each input resolution.  ...  RESULTS AND DISCUSSION As shown in CONCLUSIONS AND FUTURE WORK The presented study introduced a Convolutional Neural Network (CNN) architecture which harnesses multiple temporal resolutions to learn  ...  Single-resolution model results provided on top, multi-resolution models at the bottom. fft instance grouped instance grouped win size raw raw augmented augmented  ... 
arXiv:1811.04419v1 fatcat:dyaxepbdavduna6bdnf2rxoywa

A Multi-scale Multiple Instance Video Description Network [article]

Huijuan Xu, Subhashini Venugopalan, Vasili Ramanishka, Marcus Rohrbach, Kate Saenko
2016 arXiv   pre-print
Our paper tries to solve this problem by integrating the base CNN into several fully convolutional neural networks (FCNs) to form a multi-scale network that handles multiple receptive field sizes in the  ...  To further handle the ambiguity over multiple objects and locations, we incorporate the Multiple Instance Learning mechanism (MIL) to consider objects in different positions and at different scales simultaneously  ...  In [18] , the authors introduce the concept of multiple instance learning with FCN to make use of multi-class image labels for training, but their ultimate goal is still image segmentation.  ... 
arXiv:1505.05914v3 fatcat:chwvb7y3z5awxmhswarpwwlmsi

Multi-Segment Deep Convolution Neural Networks for Classification of Faults in Sensors at Railway Point Systems

Swati Sachan, Nishant Donchak
2019 2019 25th International Conference on Automation and Computing (ICAC)  
The second layer is fully-connected convolution layer. It extracts global temporal features.  ...  The architecture of the proposed deep learning network consists of three types of layers. The first layer is called the local convolution layer.  ...  The parameters in this multi-segment deep CNN model was trained and validated by these instances.  ... 
doi:10.23919/iconac.2019.8895081 dblp:conf/iconac/SachanD19 fatcat:g5vvcaf2h5cnvopo3auzs4ixlu

Weakly Supervised Cascaded Convolutional Networks [article]

Ali Diba, Vivek Sharma, Ali Pazandeh, Hamed Pirsiavash, Luc Van Gool
2016 arXiv   pre-print
The final stage of both architectures is a part of a convolutional neural network that performs multiple instance learning on proposals extracted in the previous stage(s).  ...  The first stage of both architectures extracts best candidate of class specific region proposals by training a fully convolutional network.  ...  The first stage extracts class specific object proposals using a fully convolutional network followed by a global average (max) pooling layer.  ... 
arXiv:1611.08258v1 fatcat:cbdt7v5m2jdr7bypq4hvec23ru

Label-PEnet: Sequential Label Propagation and Enhancement Networks for Weakly Supervised Instance Segmentation [article]

Weifeng Ge, Sheng Guo, Weilin Huang, Matthew R. Scott
2020 arXiv   pre-print
We design four cascaded modules including multi-label classification, object detection, instance refinement and instance segmentation, which are implemented sequentially by sharing the same backbone.  ...  Weakly-supervised instance segmentation aims to detect and segment object instances precisely, given imagelevel labels only.  ...  Multiple Cascaded Modules Multi-Label Classification Module.  ... 
arXiv:1910.02624v3 fatcat:c5w76g7ikrc6jgp5kzgpuh4lcu

Deep multiple instance learning for image classification and auto-annotation

Jiajun Wu, Yinan Yu, Chang Huang, Kai Yu
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
In this paper, we attempt to model deep learning in a weakly supervised learning (multiple instance learning) framework.  ...  In our setting, each image follows a dual multi-instance assumption, where its object proposals and possible text annotations can be regarded as two instance sets.  ...  As shown in Figure 2 , it contains five convolutional layers, followed by a pooling layer and three fully connected layers. We redesign the last hidden layer for multiple instance learning.  ... 
doi:10.1109/cvpr.2015.7298968 dblp:conf/cvpr/WuYHY15 fatcat:x4b4cmcuvzbdphuevi6f2upy2a

Medical Image Segmentation with 3D Convolutional Neural Networks: A Survey [article]

S Niyas, S J Pawan, M Anand Kumar, Jeny Rajan
2022 arXiv   pre-print
At present, convolutional neural networks (CNN) are the preferred choice for medical image analysis.  ...  Here, we present an extensive review of the recently evolved 3D deep learning methods in medical image segmentation.  ...  For instance, Table 1 3D CNN is due to the optimization in vector multiplication while using more parameters in a single 3D convolution kernel.  ... 
arXiv:2108.08467v3 fatcat:s2rzghycjbczpparmrflsdzujq

Control Chart Concurrent Pattern Classification Using Multi-Label Convolutional Neural Networks

Chuen-Sheng Cheng, Pei-Wen Chen, Ying Ho
2022 Applied Sciences  
The results of the study show that the recognition performance of multi-label convolutional neural network is better than traditional machine learning algorithms.  ...  Although concurrent patterns have multiple characteristics from different basic patterns, most studies have treated them as a special pattern and used the multi-class classifier to perform the classification  ...  Multi-Label Convolutional Neural Network Multi-label classification is a classification problem in which multiple target labels can be assigned to each sample (instance).  ... 
doi:10.3390/app12020787 fatcat:e5buzy7qibezxlangchkhjh4tq

Transferring CNNS to multi-instance multi-label classification on small datasets

Mingzhi Dong, Kunkun Pang, Yang Wu, Jing-Hao Xue, Timothy Hospedales, Tsukasa Ogasawara
2017 2017 IEEE International Conference on Image Processing (ICIP)  
Convolutional Neural Networks (CNNs) possess great potential to perform well on MIML tasks, since multi-level convolution and max pooling coincide with the multi-instance setting and the sharing of hidden  ...  It is typically addressed through multi-instance multi-label (MIML) classification methodologies.  ...  Hence a simple way to adapt CNNs for solving MIML problems would be to change the original multi-class classifier in the Softmax Layer into a multi-label classifier, such as the multiple binary-class logistic  ... 
doi:10.1109/icip.2017.8296498 dblp:conf/icip/DongPWXHO17 fatcat:pgsuoycqk5b47b5wsjo262h6yy

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.  ...  In multiple instance learning we find another approach closely related to our work, count-based multiple instance learning [8] .  ... 
doi:10.1109/cvprw.2015.7301276 dblp:conf/cvpr/SeguiPV15 fatcat:ackgxxegjvfubcxfy4mzwxt5fa

Evolution of Image Segmentation using Deep Convolutional Neural Network: A Survey [article]

Farhana Sultana Dept. of Computer Science, University of Gour Banga, Dept. of Computer System Sciences, Visva-Bharati University)
2020 arXiv   pre-print
In the recent era, the success of deep convolutional neural network (CNN) has influenced the field of segmentation greatly and gave us various successful models to date.  ...  In this survey, we are going to take a glance at the evolution of both semantic and instance segmentation work based on CNN.  ...  InstanceFCN: The fully convolutional network is good for single instance segmentation of an object category. But it can not distinguish multiple instances of an object.  ... 
arXiv:2001.04074v2 fatcat:jeosxdgbtzhwzjmvqqlrmancty
« Previous Showing results 1 — 15 out of 45,993 results