Filters








101 Hits in 2.8 sec

APPLICATION OF MACHINE AND DEEP LEARNING STRATEGIES FOR THE CLASSIFICATION OF HERITAGE POINT CLOUDS

E. Grilli, E. Özdemir, F. Remondino
2019 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
The classification is aimed at automatically recognizing architectural components such as columns, facades or windows in large datasets.  ...  The use of heritage point cloud for documentation and dissemination purposes is nowadays increasing.  ...  Architecture (Italy) for providing the dataset of the Neptune Temple in Paestum and Bologna's porticoes, respectively.  ... 
doi:10.5194/isprs-archives-xlii-4-w18-447-2019 fatcat:eecldwtcknaclm64zbwn7vmrty

Learning Neurosymbolic Generative Models via Program Synthesis [article]

Halley Young and Osbert Bastani and Mayur Naik
2019 arXiv   pre-print
On both synthetic and real-world data, we demonstrate that our approach is substantially better than the state-of-the-art at both generating and completing images that contain global structure.  ...  We propose to address this problem by incorporating programs representing global structure into the generative model---e.g., a 2D for-loop may represent a configuration of windows.  ...  Baseline architecture. For the baseline VAE architecture, we used a similar architecture to the PS-GM completion step (4 convolutional and 6 deconvolutional layers).  ... 
arXiv:1901.08565v1 fatcat:kz773zueh5colixcoiua4lidji

Deep Attention Neural Network for Multi-label Classification in Unmanned Aerial Vehicle Imagery

Aaliyah Alshehri, Yakoub Bazi, Nassim Ammour, Haidar Almubarak, Naif Alajlan
2019 IEEE Access  
The decoder module which is based on a long short terms memory (LSTM) network has the task of generating, in a sequential way, the classes present in the image.  ...  To tackle this issue, we propose in this paper a deep learning approach based on encoder-decoder neural network architecture with channel and spatial attention mechanisms.  ...  Architecture [26] . FIGURE 2 . 2 LSTM structure.  ... 
doi:10.1109/access.2019.2936616 fatcat:zemspkzvdnf6besuzh25xieldm

Image-based localization using LSTMs for structured feature correlation [article]

Florian Walch, Caner Hazirbas, Laura Leal-Taixé, Torsten Sattler, Sebastian Hilsenbeck, Daniel Cremers
2017 arXiv   pre-print
In this work we propose a new CNN+LSTM architecture for camera pose regression for indoor and outdoor scenes.  ...  We make use of LSTM units on the CNN output, which play the role of a structured dimensionality reduction on the feature vector, leading to drastic improvements in localization performance.  ...  To summarize, our contribution is three-fold: (i) we propose a new CNN+LSTM architecture for camera pose regression in indoor and outdoor scenes.  ... 
arXiv:1611.07890v4 fatcat:p5nnydiuzfhonjzjlp5fmmsp5e

Accurate Deep Direct Geo-Localization from Ground Imagery and Phone-Grade GPS [article]

Shaohui Sun, Ramesh Sarukkai, Jack Kwok, Vinay Shet
2018 arXiv   pre-print
Having ground truth locations for training, we are able to reach nearly lane-level accuracy.  ...  We propose a method to train a geo-spatial deep neural network (CNN+LSTM) to predict accurate geo-locations (latitude and longitude) using only ordinary ground imagery and low accuracy phone-grade GPS.  ...  In this paper, we show how CNN + LSTM architecture is capable of predicting very accurate location for navigation purposes.  ... 
arXiv:1804.07470v1 fatcat:rblygithqrh6xbxj3mq76yuwoe

Accurate Deep Direct Geo-Localization from Ground Imagery and Phone-Grade GPS

Shaohui Sun, Ramesh Sarukkai, Jack Kwok, Vinay Shet
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
Having ground truth locations for training, we are able to reach nearly lane-level accuracy.  ...  We propose a method to train a geo-spatial deep neural network (CNN+LSTM) to predict accurate geo-locations (latitude and longitude) using only ordinary ground imagery and low accuracy phone-grade GPS.  ...  In this paper, we show how CNN + LSTM architecture is capable of predicting very accurate location for navigation purposes.  ... 
doi:10.1109/cvprw.2018.00148 dblp:conf/cvpr/SunSKS18 fatcat:sanucrz63jhs7c2wdrsbytlowm

A machine learning-based framework for cost-optimal building retrofit

Chirag Deb, Zhonghao Dai, Arno Schlueter
2021 Applied Energy  
This leads to the process of exhaustive search for obtaining the cost-optimal retrofit strategy.  ...  However, these models are oblivious to the significance of the input variables for devising the retrofit strategies.  ...  The authors also thank Mario Frei for sharing information on the WSN measurements and Diego Sigrist for sharing the cost-matrix of the retrofit measures.  ... 
doi:10.1016/j.apenergy.2021.116990 fatcat:b5lgct7x4zdt5hcevigs7my7sq

Sequence to Sequence -- Video to Text [article]

Subhashini Venugopalan, Marcus Rohrbach, Jeff Donahue, Raymond Mooney, Trevor Darrell, Kate Saenko
2015 arXiv   pre-print
For this we exploit recurrent neural networks, specifically LSTMs, which have demonstrated state-of-the-art performance in image caption generation.  ...  Real-world videos often have complex dynamics; and methods for generating open-domain video descriptions should be sensitive to temporal structure and allow both input (sequence of frames) and output (  ...  Acknowledgments We thank Lisa Anne Hendricks, Matthew Hausknecht, Damian Mrowca for helpful discussions; and Anna Rohrbach for help with both movie corpora; and the  ... 
arXiv:1505.00487v3 fatcat:6jt7k5r73veobo73ue3tvm4pz4

Estimation of Heat Loss Coefficient and Thermal Demands of In-Use Building by Capturing Thermal Inertia Using LSTM Neural Networks

Martín Pensado-Mariño, Lara Febrero-Garrido, Estibaliz Pérez-Iribarren, Pablo Eguía Oller, Enrique Granada-Álvarez
2021 Energies  
Moreover, the viability of LSTM neural networks to estimate the HLC of an in-use building with an error below 4% was demonstrated.  ...  The aim of this work is to study the thermal inertia of a building developing a new methodology based on Long Short-Term Memory (LSTM) neural networks.  ...  Materials and Methods LSTM Architecture Long Short-Term Memory (LSTM) is a variety of Recurrent Neural Network (RNN) architecture created to avoid the problem of the vanishing (or exploding) gradient  ... 
doi:10.3390/en14165188 fatcat:lvgxcdrmgjdl3c763h5i5vhbne

End-to-end deep meta modelling to calibrate and optimize energy consumption and comfort [article]

Max Cohen, Gilles Nozière
2021 arXiv   pre-print
see the appendices for a complete list of these ranges.  ...  Therefore, we decided to evaluate the go-to architectures for time series: a bidirectional LSTM, a bidirectional GRU (BiGRU), a hybrid model mixing both convolutional and GRU layers (ConvGru) and a Feed  ... 
arXiv:2105.02814v2 fatcat:i3xyahaaczacdmcmymoubxh274

Improving Image-Based Localization with Deep Learning: The Impact of the Loss Function [article]

Isaac Ronald Ward, M. A. Asim K. Jalwana, Mohammed Bennamoun
2019 arXiv   pre-print
We achieve improvements in the localization accuracy of the network for indoor scenes; with decreases of up to 26.7% and 24.0% in the median positional and rotational error respectively, when compared  ...  mutually exclusive, and LSTMs with augmented loss functions is perhaps an avenue for future research.  ...  Architecture and training As stated, we primarily experiment with the PoseNet architecture.  ... 
arXiv:1905.03692v2 fatcat:csiy6e3ebfb6jg22oja465wlv4

The Image Based Positioning Technique using Inter-Pixel Relation Network

Zhenyu Zhang, Qiuling Zhao
2020 IEEE Access  
Accurate location information is very important for human beings and agents.  ...  For more information, see https://creativecommons.org/licenses/by/4.0/ VOLUME 8, 2020  ...  In [10] , a new CNN+LSTM architecture is proposed for camera pose regression in indoor and outdoor scenes.  ... 
doi:10.1109/access.2020.2992770 fatcat:hdn66acc7bfmbf666pry6fvjla

CFGs-2-NLU: Sequence-to-Sequence Learning for Mapping Utterances to Semantics and Pragmatics [article]

Adam James Summerville, James Ryan, Michael Mateas, Noah Wardrip-Fruin
2016 arXiv   pre-print
Specifically, we take a CFG authored to generate dialogue for our target application for NLU, a videogame, and train a long short-term memory (LSTM) recurrent neural network (RNN) to map the surface utterances  ...  To our knowledge, this is the first usage of seq2seq learning for conversational agents (our game's characters) who explicitly reason over semantic and pragmatic considerations.  ...  This method utilizes context-free grammars (CFGs) in conjunction with the long short-term memory (LSTM) recurrent neural network (RNN) architecture.  ... 
arXiv:1607.06852v1 fatcat:hyewo43phrh6rnmowrso75xisu

UAV Image Multi-Labeling with Data-Efficient Transformers

Laila Bashmal, Yakoub Bazi, Mohamad Mahmoud Al Rahhal, Haikel Alhichri, Naif Al Ajlan
2021 Applied Sciences  
During the training phase, we generated a second view for each image from the training set using data augmentation.  ...  In this paper, we present an approach for the multi-label classification of remote sensing images based on data-efficient transformers.  ...  However, CNN models were originally designed for single-label classification, and they cannot completely leverage the correlations among multiple classes.  ... 
doi:10.3390/app11093974 doaj:d8c45fb6d39f40d38e6ad4400a21d159 fatcat:zbmmbpnhfnd7rjevp2akf6yddu

RPNet: An End-to-End Network for Relative Camera Pose Estimation [chapter]

Sovann En, Alexis Lechervy, Frédéric Jurie
2019 Lecture Notes in Computer Science  
They also show that RPNet produces more accurate and more stable results than traditional approaches, especially for hard images (repetitive textures, textureless images, etc.).  ...  Relative pose estimation is an essential task for many computer vision problems, such as Structure from Motion (SfM), Simultaneous Localisation And Mapping (SLAM), etc.  ...  Conclusions This paper proposed a novel architecture for estimating full relative poses using an end-to-end neural network.  ... 
doi:10.1007/978-3-030-11009-3_46 fatcat:xusgbto6uvb77mkq2nuuu4pjti
« Previous Showing results 1 — 15 out of 101 results