text
stringlengths
27
153
label
stringlengths
27
153
id
int64
0
40k
infinite-dimensional approximate Finally, these we via CovDs approximations.
Finally, we approximate infinite-dimensional CovDs via these approximations.
36,400
task our of set image classification. Empirically, the to apply framework we
Empirically, we apply our framework to the task of image set classification.
36,401
for outperforms existing abnormality detection. Cardiac-MOS all work also binary
Cardiac-MOS also outperforms all existing work for binary abnormality detection.
36,402
future in potential scoring motion demonstrates clinical As first automatic the Cardiac-MOS great solution, application.
As the first automatic motion scoring solution, Cardiac-MOS demonstrates great potential in future clinical application.
36,403
Both research. available are dataset the for the and code
Both the code and the dataset are available for research.
36,404
two step follows process. a method Our
Our method follows a two step process.
36,405
a is First, trained. B domain autoencoder variational for
First, a variational autoencoder for domain B is trained.
36,406
using model Hashing our Deep Supervised we Embeddings (WDHT). name Accordingly, Tag Weakly
Accordingly, we name our model Weakly Supervised Deep Hashing using Tag Embeddings (WDHT).
36,407
mentioned. some for is work ideal that In future addition, have areas
In addition, some areas that is ideal for future work have mentioned.
36,408
the the Strength is node Received unknown trilateration. location estimated In Signal using (RSS),
In the Received Signal Strength (RSS), the unknown node location is estimated using trilateration.
36,409
attention which Specific in become coherent). is (or spatially multipath scenarios correlated for signals given
Specific attention is given for multipath scenarios in which signals become spatially correlated (or coherent).
36,410
approach and useful proposed Our offers practically into neural is interpretability. insights
Our proposed approach is practically useful and offers insights into neural interpretability.
36,411
assumption algorithms. The of typical rigidity SLAM is scene in
The assumption of scene rigidity is typical in SLAM algorithms.
36,412
scenarios monocular, dynamic DynaSLAM is for RGB-D in stereo robust configurations. and
DynaSLAM is robust in dynamic scenarios for monocular, stereo and RGB-D configurations.
36,413
monocular, our stereo in datasets. evaluate public We and RGB-D system
We evaluate our system in public monocular, stereo and RGB-D datasets.
36,414
visual DynaSLAM dynamic standard accuracy baselines outperforms the in scenarios. SLAM highly of
DynaSLAM outperforms the accuracy of standard visual SLAM baselines in highly dynamic scenarios.
36,415
small-scale achieved results of in excellent datasets. landmark problem recognition has The
The problem of landmark recognition has achieved excellent results in small-scale datasets.
36,416
of the design effectiveness the we assess waveform approach proposed Finally, by simulations. numerical
Finally, we assess the effectiveness of the proposed waveform design approach by numerical simulations.
36,417
to to add network. set samples identified neural the Bayesian using are Informative a training
Informative samples to add to the training set are identified using a Bayesian neural network.
36,418
piecewise function Mumford-Shah approximates The smooth a functional function. by a
The Mumford-Shah functional approximates a function by a piecewise smooth function.
36,419
an discretize Ambrosio-Tortorelli this Calculus via Exterior Discrete In we approximation formulation. paper, a
In this paper, we discretize an Ambrosio-Tortorelli approximation via a Discrete Exterior Calculus formulation.
36,420
of for is localization the automatic videos. endoscopic essential tool task Surgical analysis an
Surgical tool localization is an essential task for the automatic analysis of endoscopic videos.
36,421
annotated the data weak with circumvent supervision. In we to this lack work, of propose
In this work, we propose to circumvent the lack of annotated data with weak supervision.
36,422
proposed great spinal disease method clinical in diagnoses. potential has The
The proposed method has great potential in clinical spinal disease diagnoses.
36,423
becomes image an When translation is data not accessible, problem. training ill-posed paired
When paired training data is not accessible, image translation becomes an ill-posed problem.
36,424
to a single separating approach reflection an from image. present We
We present an approach to separating reflection from a single image.
36,425
separation. layer novel We loss pixel-level exclusion also that enforces a propose
We also propose a novel exclusion loss that enforces pixel-level layer separation.
36,426
foveated images, pixel image produce spatially density. with we an To rasterize varying to
To produce foveated images, we rasterize to an image with spatially varying pixel density.
36,427
set approach. of our perceptual of user A the demonstrates studies efficacy
A set of perceptual user studies demonstrates the efficacy of our approach.
36,428
the These processing task not classification jointly hand. are stages, at optimizing however,
These processing stages, however, are not jointly optimizing the classification task at hand.
36,429
of information is label into Contextual means recurrent by ReuseNet. incorporated FuseNet a version called
Contextual label information is incorporated into FuseNet by means of a recurrent version called ReuseNet.
36,430
regularization fields. and resampling and such map conditional as random pansharpening through interpolation,
pansharpening and resampling through interpolation, and map regularization such as conditional random fields.
36,431
FuseNet approaches baseline quantitative surpass qualitative in results. ReuseNet and the and both
FuseNet and ReuseNet surpass the baseline approaches in both quantitative and qualitative results.
36,432
Finally, games present suite. Atari preliminary the we results in of a subset
Finally, we present preliminary results in a subset of the Atari games suite.
36,433
enables new image This acquisition results. analysis of applications optimising possible towards best further
This further enables new applications of optimising image acquisition towards best possible analysis results.
36,434
in surveillance. It and motion capture applications has broad
It has broad applications in motion capture and surveillance.
36,435
are detecting to due and not complexity dynamic of the However, straightforward. tracking structure, human
However, due to the complexity of human dynamic structure, detecting and tracking are not straightforward.
36,436
detection Consequently, developed. different techniques have with and tracking and been different performance applications
Consequently, different detection and tracking techniques with different applications and performance have been developed.
36,437
technique a To indispensable. is fusion this problem, avoid using multi-sensor
To avoid this problem, using a multi-sensor fusion technique is indispensable.
36,438
survey detection, are In multi-sensor this presented. tracking methods of a paper, fusion comparative and
In this paper, a comparative survey of detection, tracking and multi-sensor fusion methods are presented.
36,439
as can tractography. as a direct TOMs prior input even act for or
TOMs can act as a prior or even as direct input for tractography.
36,440
the learn fully-convolutional We use encoder-decoder mapping. neural an network required architecture to
We use an encoder-decoder fully-convolutional neural network architecture to learn the required mapping.
36,441
available. data code are openly Our and our
Our code and our data are openly available.
36,442
grid outages Simulations the due reproduce behavior the of of dynamic post-contingency generators. to
Simulations reproduce the post-contingency dynamic behavior of the grid due to outages of generators.
36,443
the paper. images in self-ception Some included are
Some self-ception images are included in the paper.
36,444
numerous records from scene. field light A real-world rays light a
A light field records numerous light rays from a real-world scene.
36,445
dense devices a by light existing However, capturing time-consuming is a process. field
However, capturing a dense light field by existing devices is a time-consuming process.
36,446
significant correlation is constraint spatial dimensions. The and related between a depth most angular
The most significant constraint is a depth related correlation between angular and spatial dimensions.
36,447
method constraint employing the deep out a network. by working error-sensitive avoids neural Our
Our method avoids working out the error-sensitive constraint by employing a deep neural network.
36,448
careful fine-grained datasets. part categorization many fact, annotations In are in available
In fact, careful part annotations are available in many fine-grained categorization datasets.
36,449
is image-level of obtained aggregating region-level prediction by products element-wise The these probabilities.
The image-level prediction is obtained by aggregating element-wise products of these region-level probabilities.
36,450
proposed. To detection problem, and address this efficient YOLO were as SSD frameworks object such
To address this problem, efficient object detection frameworks such as YOLO and SSD were proposed.
36,451
improved, speed comes Although accuracy. greatly the it with price lower of a is
Although the speed is greatly improved, it comes with a price of lower accuracy.
36,452
Fire In SSD. paper, an named propose efficient SSD this we
In this paper, we propose an efficient SSD named Fire SSD.
36,453
input transformations. framework and takes sets images labeled The preprocessing predefined as of
The framework takes as input sets of labeled images and predefined preprocessing transformations.
36,454
and the the individual optimal preprocessing images. learns jointly for It transformations classifier
It jointly learns the classifier and the optimal preprocessing transformations for individual images.
36,455
computationally cheaper said version) distribution. and of an approximate
an approximate and computationally cheaper version) of said distribution.
36,456
exact MCMC sampling still accelerates the DA-MCMC distribution. applications, complex in targeting while
DA-MCMC accelerates MCMC sampling in complex applications, while still targeting the exact distribution.
36,457
a algorithm. computationally design faster, We approximate, DA-MCMC albeit
We design a computationally faster, albeit approximate, DA-MCMC algorithm.
36,458
highly dependent. problem However, acceleration the is
However, the acceleration is highly problem dependent.
36,459
antigen onto update predicted for The maps competitive receptors. scheme dynamics optimal commonly considered
The predicted optimal update scheme maps onto commonly considered competitive dynamics for antigen receptors.
36,460
labeled deep Existing large models amount of data. require network a
Existing deep network models require a large amount of labeled data.
36,461
to framework TD-GAN be can easily adapted general other The is tasks. learning and
The TD-GAN framework is general and can be easily adapted to other learning tasks.
36,462
performances. filter currently terms of (CF) based are trackers top Correlation ranked their in
Correlation filter (CF) based trackers are currently ranked top in terms of their performances.
36,463
applications. a high intensity value in diagnosis pain has assessment Automatic disease
Automatic pain intensity assessment has a high value in disease diagnosis applications.
36,464
simulation promising filtering evaluated algorithm scenario, used results. a highly with is The over commonly
The algorithm is evaluated over a commonly used filtering simulation scenario, with highly promising results.
36,465
breast cancerous poses coherency (BC) cells high to inherently significant challenges Also, cancer multi-classification. of
Also, inherently high coherency of cancerous cells poses significant challenges to breast cancer (BC) multi-classification.
36,466
investigates structural bridge of recognition video paper using the data. This components automated
This paper investigates the automated recognition of structural bridge components using video data.
36,467
is machine A train learning created video to algorithms. the simulated new dataset
A new simulated video dataset is created to train the machine learning algorithms.
36,468
video collected field. Results presented simulated as in the as for well video are data,
Results are presented for simulated video data, as well as video collected in the field.
36,469
the indispensable understanding in Scene within scene. an a component parsing is semantics
Scene parsing is an indispensable component in understanding the semantics within a scene.
36,470
on parsing. networks on based scene fully have neural Recently, new convolutional records methods achieved
Recently, methods based on fully convolutional neural networks have achieved new records on scene parsing.
36,471
task. efficiency counterpart FFT of CCIM The in that of remarkably the exceeds our
The efficiency of CCIM exceeds that of the FFT counterpart remarkably in our task.
36,472
is time applied the architecture attributes This is such first for problem. fine-grained the classification
This is the first time such architecture is applied for the fine-grained attributes classification problem.
36,473
a Convolutional with method competitive baseline. compared Neural Network We the proposed deep
We compared the proposed method with a competitive deep Convolutional Neural Network baseline.
36,474
exist practice. disadvantages there in actually two with However, GRM
However, there actually exist two disadvantages with GRM in practice.
36,475
reconstruction by the measurements randomly performances. CS cannot obtained is Another sufficient GRM that provide
Another is that the CS measurements randomly obtained by GRM cannot provide sufficient reconstruction performances.
36,476
solve STP. this an algorithm propose paper, In we the to
In this paper, we propose an algorithm to solve the STP.
36,477
for New proposed. score a vertex intensification heuristics system constructive are and purpose
New constructive heuristics and a vertex score system for intensification purpose are proposed.
36,478
benchmarks which shows on algorithm is set The encouraging of a tested results.
The algorithm is tested on a set of benchmarks which shows encouraging results.
36,479
achieve PSV evaluate result. our We method an and dataset, advanced on
We evaluate our method on PSV dataset, and achieve an advanced result.
36,480
step therefore automating workflow. the in of key cells a inspection Automated segmentation visual is
Automated segmentation of cells is therefore a key step in automating the visual inspection workflow.
36,481
have Their as anisotropy, such analyzed extensively. been spectra, discrepancy, properties or progressiveness
Their properties such as discrepancy, spectra, anisotropy, or progressiveness have been analyzed extensively.
36,482
correlations extractor can The via CNN spatial benefit component. considering also CRF from the feature
The CNN feature extractor can also benefit from considering spatial correlations via the CRF component.
36,483
Answering. propose for Question Visual paper, graph-based this we In approach a novel
In this paper, we propose a novel graph-based approach for Visual Question Answering.
36,484
by experience visuo-tactile simple body as theirs parts just stimulation. can Humans fake synchronous
Humans can experience fake body parts as theirs just by simple visuo-tactile synchronous stimulation.
36,485
example, procedurally, can or from For manually. abstracted they created measurements, generated be raw
For example, they can be abstracted from raw measurements, generated procedurally, or created manually.
36,486
consistent distributions The neighborhoods. to are various GANs and buildings synchronized style produce over
The various GANs are synchronized to produce consistent style distributions over buildings and neighborhoods.
36,487
the variability user We of the with output. provide the control over direct
We provide the user with direct control over the variability of the output.
36,488
on We large-scale examples. demonstrate several our system
We demonstrate our system on several large-scale examples.
36,489
image important patient-specific an anatomy. plays in Semantic role modeling segmentation
Semantic image segmentation plays an important role in modeling patient-specific anatomy.
36,490
and is medical both time-consuming image Manual annotation expensive.
Manual medical image annotation is both time-consuming and expensive.
36,491
time vessels to hours minutes. Kid-Net reduces segmentation from of matter kidney
Kid-Net reduces kidney vessels segmentation time from matter of hours to minutes.
36,492
and classification utilized and Feature up-sampling higher down-sampling to achieve accuracies. localization are
Feature down-sampling and up-sampling are utilized to achieve higher classification and localization accuracies.
36,493
competence. on challenging evaluation dataset show a results Quantitative qualitative and testing Kid-Net
Quantitative and qualitative evaluation results on a challenging testing dataset show Kid-Net competence.
36,494
This to cases. degradation the in leads some performance fusion
This leads to the fusion performance degradation in some cases.
36,495
features source Firstly, deep from network the extract residual is images. to used (ResNet)
Firstly, the residual network (ResNet) is used to extract deep features from source images.
36,496
maps. the and obtain is Then weight ZCA to utilized normalize features deep initial
Then ZCA is utilized to normalize the deep features and obtain initial weight maps.
36,497
using Finally, a reconstructed strategy. is image the weighted-averaging fused
Finally, the fused image is reconstructed using a weighted-averaging strategy.
36,498
integrates provide information Multi-task network neural that a jointly distinct from architectures mechanism sources.
Multi-task neural network architectures provide a mechanism that jointly integrates information from distinct sources.
36,499