Filters








1,572 Hits in 6.3 sec

Learning to Zoom: a Saliency-Based Sampling Layer for Neural Networks [article]

Adrià Recasens, Petr Kellnhofer, Simon Stent, Wojciech Matusik and Antonio Torralba
2018 arXiv   pre-print
We introduce a saliency-based distortion layer for convolutional neural networks that helps to improve the spatial sampling of input data for a given task.  ...  learns how best to sample from the underlying high resolution data in a manner which preserves task-relevant information better than uniform downsampling.  ...  We acknowledge NVIDIA Corporation for hardware donations.  ... 
arXiv:1809.03355v1 fatcat:awpyqooeprcajoiqxlzvrwnpry

Learning to Zoom: A Saliency-Based Sampling Layer for Neural Networks [chapter]

Adrià Recasens, Petr Kellnhofer, Simon Stent, Wojciech Matusik, Antonio Torralba
2018 Lecture Notes in Computer Science  
We introduce a saliency-based distortion layer for convolutional neural networks that helps to improve the spatial sampling of input data for a given task.  ...  learns how best to sample from the underlying high resolution data in a manner which preserves task-relevant information better than uniform downsampling.  ...  We acknowledge NVIDIA Corporation for hardware donations.  ... 
doi:10.1007/978-3-030-01240-3_4 fatcat:evduc4styfa7plcumsq3wyqjqm

Learning When and Where to Zoom with Deep Reinforcement Learning [article]

Burak Uzkent, Stefano Ermon
2020 arXiv   pre-print
In this direction, we propose PatchDrop a reinforcement learning approach to dynamically identify when and where to use/acquire high resolution data conditioned on the paired, cheap, low resolution images  ...  For these reasons, it is desirable to develop an automatic method to selectively use high resolution data when necessary while maintaining accuracy and reducing acquisition/run-time cost.  ...  Acknowledgements This research was supported by Stanfords Data for Development Initiative and NSF grants 1651565 and 1733686.  ... 
arXiv:2003.00425v2 fatcat:wh4bslazmngztmx3qwg66enlmu

Learning When and Where to Zoom With Deep Reinforcement Learning

Burak Uzkent, Stefano Ermon
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
In this direction, we propose PatchDrop a reinforcement learning approach to dynamically identify when and where to use/acquire high resolution data conditioned on the paired, cheap, low resolution images  ...  For these reasons, it is desirable to develop an automatic method to selectively use high resolution data when necessary while maintaining accuracy and reducing acquisition/run-time cost.  ...  Acknowledgements This research was supported by Stanford's Data for Development Initiative and NSF grants 1651565 and 1733686.  ... 
doi:10.1109/cvpr42600.2020.01236 dblp:conf/cvpr/UzkentE20 fatcat:x42vmls5pvaj7j4itxq2wqwz3i

Guided Zoom: Questioning Network Evidence for Fine-grained Classification [article]

Sarah Adel Bargal, Andrea Zunino, Vitali Petsiuk, Jianming Zhang, Kate Saenko, Vittorio Murino, Stan Sclaroff
2020 arXiv   pre-print
The reason/evidence upon which a deep convolutional neural network makes a prediction is defined to be the spatial grounding, in the pixel space, for a specific class conditional probability in the model  ...  We show that Guided Zoom improves the classification accuracy of a deep convolutional neural network model and obtains state-of-the-art results on three fine-grained classification benchmark datasets.  ...  Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright annotation thereon.  ... 
arXiv:1812.02626v2 fatcat:74dd2h3su5cnfp6xquqgefhd3e

Learning to Incorporate Texture Saliency Adaptive Attention to Image Cartoonization

Xiang Gao, Yuqi Zhang, Yingjie Tian
2022 International Conference on Machine Learning  
To this end, a novel cartoontexture-saliency-sampler (CTSS) module is proposed to dynamically sample cartoon-texturesalient patches from training data.  ...  With extensive experiments, we demonstrate that texture saliency adaptive attention in adversarial learning, as a missing ingredient of related methods in image cartoonization, is of significant importance  ...  With the thriving of deep neural networks, recent image cartoonization methods resort to learning-based framework, typically generative adversarial network (GAN) (Goodfellow et al., 2014) , to automatically  ... 
dblp:conf/icml/GaoZT22 fatcat:bjr5qzjp6nfehdsyppu5ka5nly

Learning to Incorporate Texture Saliency Adaptive Attention to Image Cartoonization [article]

Xiang Gao, Yuqi Zhang, Yingjie Tian
2022 arXiv   pre-print
To this end, a novel cartoon-texture-saliency-sampler (CTSS) module is proposed to dynamically sample cartoon-texture-salient patches from training data.  ...  With extensive experiments, we demonstrate that texture saliency adaptive attention in adversarial learning, as a missing ingredient of related methods in image cartoonization, is of significant importance  ...  With the thriving of deep neural networks, recent image cartoonization methods resort to learning-based framework, typically generative adversarial network (GAN) [5] , to automatically learn and transfer  ... 
arXiv:2208.01587v1 fatcat:mxdp45w7efaedb7qli62mkwwtq

Learning how to explain neural networks: PatternNet and PatternAttribution [article]

Pieter-Jan Kindermans, Kristof T. Schütt, Maximilian Alber, Klaus-Robert Müller, Dumitru Erhan, Been Kim, Sven Dähne
2017 arXiv   pre-print
Yet they are used on multi-layer networks with millions of parameters. This is a cause for concern since linear models are simple neural networks.  ...  DeConvNet, Guided BackProp, LRP, were invented to better understand deep neural networks. We show that these methods do not produce the theoretically correct explanation for a linear model.  ...  We are grateful to Chris Olah and Gregoire Montavon for the valuable discussions.  ... 
arXiv:1705.05598v2 fatcat:z4hno2qyk5hdpnrwa6ij74mrb4

Exploring to learn visual saliency: The RL-IAC approach [article]

Celine Craye, Timothee Lesort, David Filliat, Jean-Francois Goudou
2018 arXiv   pre-print
On the one hand, we describe a method for learning and incrementally updating a model of visual saliency from a depth-based object detector.  ...  We then demonstrate that such a saliency model learned directly on a robot outperforms several state-of-the-art saliency techniques, and that RL-IAC can drastically decrease the required time for learning  ...  ACKNOWLEDGMENT The authors would like to thank the INRIA Flowers team, and especially Pierre-Yves Oudeyer for the valuable help on the IAC aspect.  ... 
arXiv:1804.00435v1 fatcat:3gcttrpnmnf43pxvizxmxiy2p4

Deep Learning Applied to SEM Images for Supporting Marine Coralline Algae Classification

Giulia Piazza, Cecile Valsecchi, Gabriele Sottocornola
2021 Diversity  
In this work, we explored a new classification tool for coralline algae, using fine-tuning pretrained Convolutional Neural Networks (CNNs) on SEM images paired to morphological categories, including cell  ...  Our model produced promising results in terms of image classification accuracy given the constraint of a limited dataset and was tested for the identification of two ambiguous samples referred to as L.  ...  Deep learning usually refers to artificial neural networks with more than two hidden layers.  ... 
doi:10.3390/d13120640 fatcat:5gtiso2545e2vp2tlujuso7w4q

360-Degree Gaze Estimation in the Wild Using Multiple Zoom Scales [article]

Ashesh, Chu-Song Chen, Hsuan-Tien Lin
2021 arXiv   pre-print
In this work, we design a model that mimics humans' ability to estimate the gaze by aggregating from focused looks, each at a different magnification level of the face area.  ...  The model avoids the need to extract clear eye patches and at the same time addresses another important issue of face-scale variation for gaze estimation in the wild.  ...  Acknowledgement We thank the anonymous reviewers for valuable suggestions.  ... 
arXiv:2009.06924v3 fatcat:7oult55dozbozaxza5c2syx3ru

What evidence does deep learning model use to classify Skin Lesions? [article]

Xiaoxiao Li, Junyan Wu, Eric Z. Chen, Hongda Jiang
2019 arXiv   pre-print
In this paper, we propose a method to interpret the deep learning classification findings. Firstly, we propose an accurate neural network architecture to classify skin lesions.  ...  Recent years, the value of deep learning empowered computer-assisted diagnose has been shown in biomedical imaging based decision making.  ...  In this paper, we proposed a novel method based on deep convolutional neural networks and low-level image feature descriptors, which imitate clinical criteria representations, to solve skin lesion analysis  ... 
arXiv:1811.01051v3 fatcat:rg2lhdxo2ngi7gtmjpevc7q6wu

Weakly Supervised Learning Guided by Activation Mapping Applied to a Novel Citrus Pest Benchmark [article]

Edson Bollis, Helio Pedrini, Sandra Avila
2020 arXiv   pre-print
In this context, we design a weakly supervised learning process guided by saliency maps to automatically select regions of interest in the images, significantly reducing the annotation task.  ...  In addition, we create a large citrus pest benchmark composed of positive samples (six classes of mite species) and negative samples.  ...  [33] proposed a weakly supervised CNN framework, called Multiple Instance Learning Convolutional Neural Networks (MILCNN), which fuses residual network and multiple instances learning loss layer.  ... 
arXiv:2004.11252v1 fatcat:dvv3rlj2zjfwnkjvyyfhdxhqla

Pitch Estimation Of Choir Music Using Deep Learning Strategies: From Solo To Unison Recordings

Helena Cuesta
2017 Zenodo  
Then, we train several deep learning architectures to extract pitch infor- mation from monophonic singing voice signals, and adapt them afterwards to model unison performances.  ...  The presented models provide a first step towards de automatic transcription of choir singing recordings, and the unison model is a useful resource for choir singing synthesis.  ...  To do so, we decided to train a neural network based on the ones we used for the monophonic pitch estimation and adapt it to predict these two values.  ... 
doi:10.5281/zenodo.1108524 fatcat:ybmgeomdy5fn5nm5j7vgxw2zf4

CAMOU: Learning Physical Vehicle Camouflages to Adversarially Attack Detectors in the Wild

Yang Zhang, Hassan Foroosh, Philip David, Boqing Gong
2019 International Conference on Learning Representations  
In particular, we learn a camouflage pattern to hide vehicles from being detected by state-of-the-art convolutional neural network based detectors. Our approach alternates between two threads.  ...  In the first, we train a neural approximation function to imitate how a simulator applies a camouflage to vehicles and how a vehicle detector performs given images of the camouflaged vehicles.  ...  ACKNOWLEDGMENT This work was in part supported by the NSF grants IIS-1212948, IIS-1566511, and a gift from Uber Technologies Inc.  ... 
dblp:conf/iclr/ZhangFDG19 fatcat:sl2zumsvunejzcyf4422gsjrfe
« Previous Showing results 1 — 15 out of 1,572 results