Filters








922 Hits in 5.6 sec

DriveGuard: Robustification of Automated Driving Systems with Deep Spatio-Temporal Convolutional Autoencoder [article]

Andreas Papachristodoulou, Christos Kyrkou, Theocharis Theocharides
2021 arXiv   pre-print
In order to mitigate such phenomena, we propose DriveGuard, a lightweight spatio-temporal autoencoder, as a solution to robustify the image segmentation process for autonomous vehicles.  ...  By first processing camera images with DriveGuard, we offer a more universal solution than having to re-train each perception model with noisy input.  ...  supported by the European Union's Horizon 2020 research and innovation programme under grant agreement No 739551 (KIOS CoE) and from the Government of the Republic of Cyprus through the Directorate General for  ... 
arXiv:2111.03480v1 fatcat:uygberuoynbrtcuzjujrld5fim

Cancer Diagnosis Using Deep Learning: A Bibliographic Review

Khushboo Munir, Hassan Elahi, Afsheen Ayub, Fabrizio Frezza, Antonello Rizzi
2019 Cancers  
Boltzmann's machine (RBM), stacked autoencoders (SAE), convolutional autoencoders (CAE), recurrent neural networks (RNNs), long short-term memory (LTSM), multi-scale convolutional neural network (M-CNN  ...  In particular, deep neural networks can be successfully used for intelligent image analysis.  ...  They used 2D CT images for pulmonary node classifications, whereas, in [61] , researchers used 3D CT images on multi-view CNN, which were could be used for end-to-end training.  ... 
doi:10.3390/cancers11091235 pmid:31450799 pmcid:PMC6770116 fatcat:ktuuttdu6zc7phj3mahp5yynxq

A Deep Learning Approach to Object Affordance Segmentation

Spyridon Thermos, Petros Daras, Gerasimos Potamianos
2020 ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
In particular, we design an autoencoder that is trained using ground-truth labels of only the last frame of the sequence, and is able to infer pixel-wise affordance labels in both videos and static images  ...  We show that our model achieves competitive results compared to strongly supervised methods on SOR3D-AFF, while being able to predict affordances for similar unseen objects in two affordance image-only  ...  An end-to-end autoencoder, trained using sequences of RGB-D data, learns to predict pixel-wise affordance labels, while an attention mechanism placed at the latent space of the model is responsible for  ... 
doi:10.1109/icassp40776.2020.9054167 dblp:conf/icassp/ThermosDP20 fatcat:vz3h4jzwj5a7daufhvik2elxjy

DeepHealth: Review and challenges of artificial intelligence in health informatics [article]

Gloria Hyunjung Kwak, Pan Hui
2020 arXiv   pre-print
practical use.  ...  The demand for it in health informatics is also increasing, and we can expect to see the potential benefits of its applications in healthcare.  ...  And then, they concatenated the last layers for final prediction. In [129] and [137] , both used multi-scale CNN architecture in multiple streams for pulmonary nodule classification.  ... 
arXiv:1909.00384v2 fatcat:sy7pm2c2uvdd3pal2russn4xri

An unsupervised approach to Geographical Knowledge Discovery using street level and street network images [article]

Stephen Law, Mateo Neira
2019 arXiv   pre-print
The approach allows for meaningful explanations using a combination of geographical and generative visualisations to explore the latent space, and to show how the learned representation can be used to  ...  This research contributes to the ladder, where we show how latent variables learned from unsupervised learning methods on urbanimages can be used for geographic knowledge discovery.  ...  We will then use these components for down stream tasks such as prediction and classification. The process can be seen in figure 1.  ... 
arXiv:1906.11907v2 fatcat:rs6xpkyxqjbade4g3mli5im35a

A Deep Evaluator for Image Retargeting Quality by Geometrical and Contextual Interaction

Bin Jiang, Jiachen Yang, Qinggang Meng, Baihua Li, Wen Lu
2018 IEEE Transactions on Cybernetics  
Training Data Considering Segmented Stacked AutoEncoder In [47] , SAE is proved to outperform DBN in a particular case.  ...  On the other hand, stacked AutoEnCoder makes use of greedy training method layer by layer to train each layer of the network sequentially.  ... 
doi:10.1109/tcyb.2018.2864158 pmid:30183651 fatcat:ocjk7guh2jgb5kn7xp6mznvcjq

Multi-modal Sentiment Classification with Independent and Interactive Knowledge via Semi-supervised Learning

Dong Zhang, Shoushan Li, Qiaoming Zhu, Guodong Zhou
2020 IEEE Access  
The key idea is to leverage the semi-supervised variational autoencoders to mine more information from unlabeled data for multi-modal sentiment analysis.  ...  In this paper, we aim to reduce the annotation effort for multi-modal sentiment classification via semi-supervised learning.  ...  We use the implementation by [19] . And we also implement the Stacked LSTM, Bidirectional LSTM and Stacked Bidirectional LSTMs for stronger baselines.  ... 
doi:10.1109/access.2020.2969205 fatcat:bkcydvpc7bgs3l6n4usfz7lvcy

A Continuous Learning Framework for Activity Recognition Using Deep Hybrid Feature Models

Mahmudul Hasan, Amit K. Roy-Chowdhury
2015 IEEE transactions on multimedia  
Given the segmented activities from streaming videos, we learn features in an unsupervised manner using deep hybrid networks, which have the ability to take the advantage of both the local hand-engineered  ...  Additionally, we use active learning to train the activity classifier using a reduced amount of manually labeled instances.  ...  Some recent works used active learning in several computer vision related applications such as streaming data [28] , image segmentation [29] , image and object classification [30] , video recognition  ... 
doi:10.1109/tmm.2015.2477242 fatcat:ve55gpwl3zaljcgsxgvyo6jfwy

Dual Supervised Autoencoder Based Trajectory Classification Using Enhanced Spatio-Temporal Information

Sichong Lu, Ying Xia
2020 IEEE Access  
[9] proposed a method to generate trajectory images, in particular, they transformed the trajectory segments into grid images and learned high-level features from them using an autoencoder.  ...  AUTOENCODER Autoencoder was first proposed in [45] for the pretraining stage of a neural network, which is a feed-forward multi-layer neural network.  ... 
doi:10.1109/access.2020.3026110 fatcat:rckw43yesjfpvfizj25clkummu

Multivariate Time Series as Images: Imputation Using Convolutional Denoising Autoencoder

Abdullah Al Safi, Christian Beyer, Vishnu Unnikrishnan, Myra Spiliopoulou
2020 International Symposium on Intelligent Data Analysis  
2D images using Gramian Angular Summation Field (GASF).  ...  This work proposes a modified Convolutional Denoising Autoencoder (CDA) based approach to impute multivariate time series data in combination with a preprocessing step that encodes time series data into  ...  This work is partially funded by the German Research Foundation, project OSCAR "Opinion Stream Classification with Ensembles and Active Learners".  ... 
doi:10.1007/978-3-030-44584-3_1 dblp:conf/ida/SafiBUS20 fatcat:ohmm3kxiyzgwth7od2t6grlvp4

Skeleton Based Action Recognition using a Stacked Denoising Autoencoder with Constraints of Privileged Information [article]

Zhize Wu, Thomas Weise, Le Zou, Fei Sun, Ming Tan
2020 arXiv   pre-print
Based on the concept of learning under privileged information, we integrate action categories and temporal coordinates into a stacked denoising autoencoder in the training phase, to preserve category and  ...  We finally represent the sequences using a Fourier Temporal Pyramid (FTP) representation and perform classification using a combination of LWSR registration, FTP representation, and a linear Support Vector  ...  Based on such color images, a multi-stream CNN-based model is applied to extract and fuse deep features.  ... 
arXiv:2003.05684v1 fatcat:2ea7eqvzprd27ikumoquxeoo5y

Deep Learning Meets Hyperspectral Image Analysis: A Multidisciplinary Review

Alberto Signoroni, Mattia Savardi, Annalisa Baronio, Sergio Benini
2019 Journal of Imaging  
Deep learning approaches certainly offer a great variety of opportunities for solving classical imaging tasks and also for approaching new stimulating problems in the spatial–spectral domain.  ...  architectures to solve specific tasks in different application fields.  ...  Autoencoders and Deep Belief Networks Autoencoders (AEs) and Stacked Autoencoders (SAEs) have been widely used in hyperspectral imagery for different tasks, mainly in RS but also in food-quality applications  ... 
doi:10.3390/jimaging5050052 pmid:34460490 fatcat:ledlmt42bfdtdhe7tvj2dl2rwm

DSNet: A Dual-Stream Framework for Weakly-Supervised Gigapixel Pathology Image Analysis [article]

Tiange Xiang, Yang Song, Chaoyi Zhang, Dongnan Liu, Mei Chen, Fan Zhang, Heng Huang, Lauren O'Donnell, Weidong Cai
2022 arXiv   pre-print
The WSI label is then predicted with a Dual-Stream Network (DSNet), which takes the transformed local patch embeddings and multi-scale thumbnail images as inputs and can be trained by the image-level label  ...  With image-level labels only, patch-wise classification would be sub-optimal due to inconsistency between the patch appearance and image-level label.  ...  As the name suggests, our DSNet processes the two matrices T and V in two separate streams with stacks of the aforementioned multi-scale blocks and concurrent bottleneck blocks.  ... 
arXiv:2109.05788v2 fatcat:jbuftlyxijhyhfp7n4266a6laa

Safe Robot Navigation via Multi-Modal Anomaly Detection

Lorenz Wellhausen, Rene Ranftl, Marco Hutter
2020 IEEE Robotics and Automation Letters  
We propose to overcome this issue by using anomaly detection on multi-modal images for traversability classification, which is easily scalable by training in a self-supervised fashion from robot experience  ...  In this work, we evaluate multiple anomaly detection methods with a combination of uni- and multi-modal images in their performance on data from different environmental conditions.  ...  We hereafter refer to images as the stack of RGB and depth images and potentially derived quantities. B.  ... 
doi:10.1109/lra.2020.2967706 fatcat:7t5acmueorff7flufz6wvmtnru

Semantic Segmentation of Earth Observation Data Using Multimodal and Multi-scale Deep Networks [article]

Nicolas Audebert , Sébastien Lefèvre
2016 arXiv   pre-print
Our contributions are the following: 1) we transfer efficiently a DFCNN from generic everyday images to remote sensing images; 2) we introduce a multi-kernel convolutional layer for fast aggregation of  ...  Especially, we train a variant of the SegNet architecture on remote sensing data over an urban area and study different strategies for performing accurate semantic segmentation.  ...  The Vaihingen data set was provided by the German Society for Photogrammetry, Remote Sensing and Geoinformation (DGPF) [39] : http://www.ifp.uni-stuttgart.de/dgpf/DKEP-Allg.html.  ... 
arXiv:1609.06846v1 fatcat:7tu6as23pbd7xgsnjzokor3bp4
« Previous Showing results 1 — 15 out of 922 results