text
stringlengths
27
153
label
stringlengths
27
153
id
int64
0
40k
producing function of capable Leveraging is high-quality this a output. as guide, input color edge-aware
Leveraging color input as a guide, this function is capable of producing high-quality edge-aware output.
37,800
Unlike manual works, matching of point clusters. existing require not we do annotation many
Unlike many existing works, we do not require manual annotation of matching point clusters.
37,801
this set-based The i.e. of face is work recognition, objective
The objective of this work is set-based face recognition, i.e.
37,802
implicitly these Both of set-wise classification. for during training learnt qualities are
Both of these qualities are learnt implicitly during training for set-wise classification.
37,803
these this In paper, problem explore the relationship we domains. two between
In this paper, we explore the relationship between these two problem domains.
37,804
capacity. the discussed is the system performance average of proposed of in terms hybrid The
The performance of the proposed hybrid system is discussed in terms of the average capacity.
37,805
derived by analytical are Monte validated The simulations. Carlo expressions
The derived analytical expressions are validated by Monte Carlo simulations.
37,806
of a is also provided. hybrid/PLC between A the hybrid comparison and parallel performance systems
A comparison between the performance of a parallel hybrid/PLC and hybrid systems is also provided.
37,807
hybrid one. It found that the hybrid/PLC system the outperforms is
It is found that the hybrid/PLC system outperforms the hybrid one.
37,808
attention. a lot Recently, style has of received transfer
Recently, style transfer has received a lot of attention.
37,809
at intersections for intentions the drivers Understanding is autonomous critical vehicles. component a of
Understanding the intentions of drivers at intersections is a critical component for autonomous vehicles.
37,810
we give paper contribution. this two In
In this paper we give two contribution.
37,811
affects it system. such condition Epilepsy is a brain the neurological the nervous and that
Epilepsy is a neurological condition such that it affects the brain and the nervous system.
37,812
great seizure patients. has of importance clinical epileptic in therapy Hence, identification
Hence, seizure identification has great importance in clinical therapy of epileptic patients.
37,813
results higher than accuracies quite are published. These earlier
These accuracies are quite higher than earlier results published.
37,814
models used framework. train and the scans Then were annotations developed of to segmentation these
Then these scans and annotations were used to train segmentation models of the developed framework.
37,815
attacks into Adversarial find models fool can images. misclassifying perturbations that
Adversarial attacks find perturbations that can fool models into misclassifying images.
37,816
simulate lighting can natural In and shadows. find addition, perturbations natural we phenomena harmonic like
In addition, we find harmonic perturbations can simulate natural phenomena like natural lighting and shadows.
37,817
the neighbor image addresses of This nearest applied large-scale retrieval. approximate domain in search work
This work addresses approximate nearest neighbor search applied in the domain of large-scale image retrieval.
37,818
off-line construction the we an the propose of framework testing Within search structures. efficient group
Within the group testing framework we propose an efficient off-line construction of the search structures.
37,819
embedding combined methods. method existing be can The with naturally
The method can be naturally combined with existing embedding methods.
37,820
We evaluate the databases. on multiple face anti-spoofing proposed method
We evaluate the proposed method on multiple face anti-spoofing databases.
37,821
results due The spoof improvements promising to modeling. show our noise
The results show promising improvements due to our spoof noise modeling.
37,822
Deep many achievement in has remarkable learning made fields.
Deep learning has made remarkable achievement in many fields.
37,823
usually of the amount of parameters demands networks a neural However, data. large labeled learning
However, learning the parameters of neural networks usually demands a large amount of labeled data.
37,824
called few-shot task is learning. This specific
This specific task is called few-shot learning.
37,825
The the measurement of scatter. volume of for class used is simplex the
The volume of the simplex is used for the measurement of class scatter.
37,826
learning. verify effectiveness miniImageNet simplex few-shot the of algorithm Omniglot our on Experiments on and
Experiments on Omniglot and miniImageNet verify the effectiveness of our simplex algorithm on few-shot learning.
37,827
variability this of examine the across objects different visual paper, i.e. ad categories, we In
In this paper, we examine the visual variability of objects across different ad categories, i.e.
37,828
persuasive. causes visually advertisement an be to what
what causes an advertisement to be visually persuasive.
37,829
properties are these derived. filters of The
The properties of these filters are derived.
37,830
to their we grouped be particular frequency their outputs response. can show according that In
In particular we show that their outputs can be grouped according to their frequency response.
37,831
understanding. an surveillance camera task is scene important for network a tracking Target in and
Target tracking in a camera network is an important task for surveillance and scene understanding.
37,832
views illumination and task variation challenging due disjoint in cameras. different is to The
The task is challenging due to disjoint views and illumination variation in different cameras.
37,833
this In features. were proposed using graph-based many direction, methods appearance-based
In this direction, many graph-based methods were proposed using appearance-based features.
37,834
illumination variation However, in the FOVs. high appearance information camera fades different the with
However, the appearance information fades with high illumination variation in the different camera FOVs.
37,835
will network the that policy topology. learns camera We the show
We will show that the policy learns the camera network topology.
37,836
the on NLPR of the dataset. proposed the MCT We performance method demonstrate
We demonstrate the performance of the proposed method on the NLPR MCT dataset.
37,837
advantages and of discuss we Finally, the various disadvantages schemes.
Finally, we discuss advantages and disadvantages of the various schemes.
37,838
mesh methods Markov the over as State-of-the-art Random model a Field facets. labeling
State-of-the-art methods model mesh labeling as a Markov Random Field over the facets.
37,839
compared the our compared We Matlab-toolbox Intlab. with results have
We have compared compared our results with the Matlab-toolbox Intlab.
37,840
have producing narrower superior that intervals. our is shown experiments method Numerical
Numerical experiments have shown that our method is superior producing narrower intervals.
37,841
of have principles, a been alternative supplemented DP the number postulates, ways. in These
These principles, the DP postulates, have been supplemented in a number of alternative ways.
37,842
criticised position unacceptably this has However, as recently been being strong.
However, this position has recently been criticised as being unacceptably strong.
37,843
privacy-preserving data variables These are mining for latent applications. features as proposed
These latent variables are proposed as features for privacy-preserving data mining applications.
37,844
of the we problem paper, domain In this consider adaptation.
In this paper, we consider the problem of domain adaptation.
37,845
In using on results and data. simulations real-measured addition, demonstrate experimental benefits these we
In addition, we demonstrate these benefits using experimental results on simulations and real-measured data.
37,846
make for on explore option two based contributions. methods We and algorithmic discovery variational inference
We explore methods for option discovery based on variational inference and make two algorithmic contributions.
37,847
in representations task action frame in sequences role the play important recognition. Spatio-temporal an of
Spatio-temporal representations in frame sequences play an important role in the task of action recognition.
37,848
expensive flow) However, requires framework. and an has two-stream (RGB optical computational and it cost
However, it has an expensive computational cost and requires two-stream (RGB and optical flow) framework.
37,849
downstream performance improvements. addition, deep integrated SSNs into easily resulting can in networks In be
In addition, SSNs can be easily integrated into downstream deep networks resulting in performance improvements.
37,850
artifacts to approach a these of address propose sources two-stage, all deep-learning of simultaneously. We
We propose a two-stage, deep-learning approach to address all of these sources of artifacts simultaneously.
37,851
their a systems size, challenge. laser-based communication integrating high-throughput however, remains with Due to them
Due to their size, however, integrating them with high-throughput laser-based communication systems remains a challenge.
37,852
method proposed demonstrate experimental state-of-the-art the outperforms that matching our graph results The algorithms.
The experimental results demonstrate that our proposed method outperforms the state-of-the-art graph matching algorithms.
37,853
signature technique verification considers a handwritten biometric. authentication that is an Signature as
Signature verification is an authentication technique that considers handwritten signature as a biometric.
37,854
to dimension the Furthermore reduction is feature vectors. technique applied large
Furthermore dimension reduction technique is applied to the large feature vectors.
37,855
system is the support and vector classifier model using machine evaluated trained algorithm. (SVM) A
A system model is trained and evaluated using the support vector machine (SVM) classifier algorithm.
37,856
introduced is of optical EEG. In flow variant information represent the to addition,
In addition, optical flow is introduced to represent the variant information of EEG.
37,857
generated available. directly The proportional the power is wind to speed
The power generated is directly proportional to the wind speed available.
37,858
been applied Neural core have to Convolutional Networks the age-related (CNN) research as framework.
Convolutional Neural Networks (CNN) have been applied to age-related research as the core framework.
37,859
CNN Modern are based based implementations. algebra typically point linear floating on
Modern CNN are typically based on floating point linear algebra based implementations.
37,860
constrained in environments. important power particularly is compute This
This is particularly important in power constrained compute environments.
37,861
formulate Then, we representation quantitatively hardware between data the and efficiency. relationship
Then, we quantitatively formulate the relationship between data representation and hardware efficiency.
37,862
that are the It models responses same. of appeared the almost both
It appeared that the responses of both models are almost the same.
37,863
this to we issues. present a address In model novel these paper,
In this paper, we present a novel model to address these issues.
37,864
gaze the and particular, handles represents In model explicitly gaze direction out-of-frame our targets.
In particular, our model explicitly represents the gaze direction and handles out-of-frame gaze targets.
37,865
different multi-task using leverage a datasets three We approach. learning
We leverage three different datasets using a multi-task learning approach.
37,866
methods the count use of to learn regression-based interest. explicitly optimization where State-of-the-art they objects
State-of-the-art methods use regression-based optimization where they explicitly learn to count the objects of interest.
37,867
capture linearity, Due non-linear they can extreme and to not expressions. this deformations
Due to this linearity, they can not capture extreme deformations and non-linear expressions.
37,868
localization activity literature the in Most methods annotation from suffer frame-wise burden requirement. of the
Most activity localization methods in the literature suffer from the burden of frame-wise annotation requirement.
37,869
a may labels effort. labeling reducing such from weak solution be manual potential towards Learning
Learning from weak labels may be a potential solution towards reducing such manual labeling effort.
37,870
basic signal. the frequently set smallest Noise on limit the level
Noise level frequently set the basic limit on the smallest signal.
37,871
always RF technology. in Low play noise amplifiers role a significant
Low noise amplifiers always play a significant role in RF technology.
37,872
with The been the single-gate LNA. has done comparison
The comparison has been done with the single-gate LNA.
37,873
understanding deep networks computations. features in their can neural (DNNs) help Visualizing
Visualizing features in deep neural networks (DNNs) can help understanding their computations.
37,874
are statistical widely among most and used simplest Permutation tests the tools.
Permutation tests are among the simplest and most widely used statistical tools.
37,875
be p-values computed Their permutations. sampling by a can straightforward of
Their p-values can be computed by a straightforward sampling of permutations.
37,876
The asymptotically than algorithm practically sampling. is straightforward and faster
The algorithm is practically and asymptotically faster than straightforward sampling.
37,877
we the tackle In of work, counting images. crowd this in problem
In this work, we tackle the problem of crowd counting in images.
37,878
based density Convolutional to estimation (CNN) We Neural a this approach problem. Network present solve
We present a Convolutional Neural Network (CNN) based density estimation approach to solve this problem.
37,879
high resolution in is map go a challenging task. one density a Predicting
Predicting a high resolution density map in one go is a challenging task.
37,880
We present matting. the (GAN) first natural image adversarial for generative network
We present the first generative adversarial network (GAN) for natural image matting.
37,881
features to obtain performance. multiple better natural it's Therefore, together integrate to
Therefore, it's natural to integrate multiple features together to obtain better performance.
37,882
its utilizing sample nearest neighbors. by k MRPE each reconstructs
MRPE reconstructs each sample by utilizing its k nearest neighbors.
37,883
segmentation offers quality). improved ratio, slightly quantitative results signal-to-noise Furthermore, also it (peak
Furthermore, it also offers slightly improved quantitative results (peak signal-to-noise ratio, segmentation quality).
37,884
of number the Another advantage presented of is technique the hyperparameters. low
Another advantage of the presented technique is the low number of hyperparameters.
37,885
effectiveness tasks results Superior methods the on demonstrate approach. of over state-of-the-art our both the
Superior results over the state-of-the-art methods on both tasks demonstrate the effectiveness of our approach.
37,886
group by clustering a of The up such method a constructs non-local spatial-spectrum cubes. series
The method constructs such a group by clustering up a series of non-local spatial-spectrum cubes.
37,887
adopted NLCTF solve is the strategy the to Finally, model. split-Bregman
Finally, the split-Bregman strategy is adopted to solve the NLCTF model.
37,888
The that outperforms method NLCTF results show competitors. other the the
The results show that the NLCTF method outperforms the other competitors.
37,889
in networks. paper traffic deep for road a proposes approach complex flow prediction learning This
This paper proposes a deep learning approach for traffic flow prediction in complex road networks.
37,890
by a pooling replaces neural we that routing. network max dynamic Thus, propose capsules with
Thus, we propose a neural network with capsules that replaces max pooling by dynamic routing.
37,891
gaze only human from a images natural task. Estimating eye is challenging
Estimating human gaze from natural eye images only is a challenging task.
37,892
problem. gaze ill-posed achieving is an highly Hence, accurate estimates
Hence, achieving highly accurate gaze estimates is an ill-posed problem.
37,893
to localization reduce and and the errors. strategies size, relative the instance absolute
the instance absolute and relative size, and strategies to reduce the localization errors.
37,894
AVA for localization simple dataset. action baseline the on We introduce a
We introduce a simple baseline for action localization on the AVA dataset.
37,895
contents a images. vision very the powerful from method for the understanding Computer is
Computer vision is a very powerful method for understanding the contents from the images.
37,896
sheep. weight to measuring the enabled us of human has involvement in the minimize It
It has enabled us to minimize the human involvement in measuring weight of the sheep.
37,897
starting model. (up-down) modular point is bottom-up Our re-implementation of the a top-down
Our starting point is a modular re-implementation of the bottom-up top-down (up-down) model.
37,898
performs however It dataset. to are particular that supervised fully ones, comparably a fine-tuned to
It performs comparably to fully supervised ones, that however are fine-tuned to a particular dataset.
37,899