In this work, we introduce an anisotropic minimal path model based on a new Riemannian tensor integrating the crossing-adaptive anisotropic radius-lifted tensor field and the front freezing indicator by appearance and path features. The non-local path feature only can be obtained during the geodesic distance computation process by the fast marching method. The predefined criterion derived from path

Sparsity preserving projection (SPP), as a widely used linear unsupervised dimensionality reduction (DR) method, is designed to preserve the sparse reconstructive relationship of the raw data. SPP constructs an affinity weight matrix by solving a sparse representation model which does not need any parameters. Moreover, the obtained projection may contain some discriminating information even if no prior

The promising achievement of sparse ranking in image-based recognition gives rise to a number of development on person re-identification (Re-ID) which aims to reconstruct the probe as a linear combination of few atoms/images from an over-complete dictionary/gallery. However, most of the existing sparse ranking based Re-ID methods lack considering the geometric relationships between probe, gallery,

The paper addresses two issues relative to machine learning on 2D+X data volumes, where 2D refers to image observation and X denotes a variable that can be associated with time, depth, wavelength, etc.. The first issue addressed is conditioning these structured volumes for compatibility with respect to convolutional neural networks operating on 2D image file formats. The second issue is associated

With the development of social media and smartphones, people share their daily lives via a large number of images, but the convince also raises a problem of privacy leakage. Therefore, effective methods are needed to infer the privacy risk of images and identify images that may disclose privacy. Several works have tried to solve this problem with deep learning models. However, we know little about

Image-text matching has drawn much attention recently with the rapid growth of multi-modal data. Many effective approaches have been proposed to solve this challenging problem, but limited effort has been devoted to re-ranking methods. Compared with the uni-modal re-ranking methods, modality heterogeneity is the major difficulty when designing a re-ranking method in the cross-modal field, which mainly

Although Convolutional Neural Networks (CNNs) have made substantial improvements in many computer vision tasks, there remains room for improvements in image-based action recognition due to the limited capability to exploit the body structure information.In this work, we propose a unified deep model to explicitly explore body structure information and fuse multiple body structure cues for robust action

We propose a new point set registration method based on mixture framework and variational inference. A three-phase registration strategy (TRS) is proposed to automatically process point set registration problem in different cases. A Gaussian variational mixture model (GVMM) with isotropic and anisotropic components under the variational inference framework is designed to weaken the effect of outliers

This paper addresses the person search task, which is a computer vision technology that finds the location of a pedestrian and retrieves it in a video taken by a single camera or multiple cameras. This task is much more challenging than the conventional settings for person re-identification or pedestrian detection since the search is susceptible to factors such as different resolutions, similar pedestrians

Network embedding plays a pivotal role in network analysis, due to the capability of encoding each node to a low-dimensional dense feature vector. However, most existing network embedding approaches only focus on preserving structural information in the network. The text features and category attributes of nodes are ignored, which are important to network analysis. In this paper, we propose an innovative

In this paper, we propose a novel Deep Kinship Matching and Recognition (DKMR) framework for multi-person kinship matching and recognition, which is a complicated and challenging task with little previous literature. Compared with most existing kinship understanding methods that mainly work on matching kinship in pairwise face images, we target at recognizing the exact kinship in nuclear family photos

Multi-label feature selection plays an indispensable role in multi-label learning, which eliminates irrelevant and redundant features while retaining relevant features. Most of existing multi-label feature selection methods employ two strategies to construct feature selection models: extracting label correlations to guide feature selection process and maintaining the consistency between the feature

Existing tracking-by-detection approaches build trackers on binary classifiers. Despite achieving state-of-the-art performance on tracking benchmarks, these trackers pay limited attention to data imbalance issue, e.g, positive and negative, easy and hard. In this paper, we demonstrate that separately learning feature embeddings corresponding to negative samples with different semantic characteristics

Video salient object detection is a challenging and important problem in computer vision domain. In recent years, deep-learning based methods have contributed to significant improvements in this domain. This paper provides an overview of recent developments in this domain and compares the corresponding methods up to date, including 1) classification of the state-of-the-art methods and their frameworks;

Developing tools to understand and visualize lifestyle is of high interest when addressing the improvement of habits and well-being of people. Routine, defined as the usual things that a person does daily, helps describe the individuals’ lifestyle. With this paper, we are the first ones to address the development of novel tools for automatic discovery of routine days of an individual from his/her egocentric

Cross-modal retrieval aims to realize accurate and flexible retrieval across different modalities of data, e.g., image and text, which has achieved significant progress in recent years, especially since generative adversarial networks (GAN) were used. However, there still exists much room for improvement. How to jointly extract and utilize both the modality-specific (complementarity) and modality-shared

Equipped with powerful convolutional neural networks (CNNs), generative models have achieved tremendous success in various vision applications. However, deep generative networks suffer from high computational and memory costs in both model training and deployment. While many efforts have been devoted to accelerate discriminative models by quantization, effectively reducing the costs for deep generative

Large-scale datasets play a fundamental role in training deep learning models. However, dataset collection is difficult in domains that involve sensitive information. Collaborative learning techniques provide a privacy-preserving solution, by enabling training over a number of private datasets that are not shared by their owners. However, recently, it has been shown that the existing collaborative

Recognising and locating image patches or sets of image features is an important task underlying much work in computer vision. Traditionally this has been accomplished using template matching. However, template matching is notoriously brittle in the face of changes in appearance caused by, for example, variations in viewpoint, partial occlusion, and non-rigid deformations. This article tests a method

Image representation is the key factor influencing the accuracy of remote sensing image segmentation. Traditional algorithms rely on the pixel-wise characteristics exhibited in the feature space. They introduce spatial information by establishing the connections between neighboring pixels in the neighborhood system. But the spectral-spatial features cannot be well expressed. In this paper, a Riemannian

We study the problem of clustering validation, i.e., clustering evaluation without knowledge of ground-truth labels, for the increasingly-popular framework known as subspace clustering. Existing clustering quality metrics (CQMs) rely heavily on a notion of distance between points, but common metrics fail to capture the geometry of subspace clustering. We propose a novel point-to-point pseudometric

Recent progress on one-stage detectors focuses on improving the quality of bounding boxes, while they pay less attention to the classification head. In this work, we focus on investigating the influence of the classification head. To understand the behavior of the classifier in one-stage detectors, we resort to the methods of the Explainable Deep Learning area. We visualize its learned representations

Many convolutional neural network (CNN)-based approaches for stereoscopic salient object detection involve fusing either low-level or high-level features from the color and disparity channels. The former method generally produces incomplete objects, whereas the latter tends to blur object boundaries. In this paper, a coupled CNN (CoCNN) is proposed to fuse color and disparity features from low to high

Most existing sparse representation-based (SR) fusion methods consider the local information of each image patch independently during fusion. Some spatial artifacts are easily introduced to the fused image. A sliding window technology is often employed by these methods to overcome this issue. However, this comes at the cost of high computational complexity. Alternatively, we come up with a novel multi-focus

Decision trees are commonly used for learning and extracting classification rules from data. The fuzzy rule based decision tree (FRDT) is very representative owing to its better robustness and generalization. However, FRDT cannot work well on the analysis of large-scale data sets. One solution for this problem is parallel computing. A proved effective parallel computing model is Map-Reduce. Ensemble

Matching partially overlapping point sets is a challenging problem in computer vision. To achieve this goal, we model point matching as a mixed linear assignment - least square problem. By eliminating the transformation variable, we reduce the minimization problem to a concave optimization problem with the property that the objective function can be converted into a form with few nonlinear terms. We

We propose an end-to-end ensemble method for person re-identification (ReID) to address the problem of overfitting in discriminative models. These models are known to converge easily, but they are biased to the training data in general and may produce a high model variance, which is known as overfitting. The ReID task is more prone to this problem due to the large discrepancy between training and test

Accurately detecting scene text is a challenging task due to perspective distortion, scale variance, varied orientations, uneven illumination. Among them, scale variance has always been a core issue and generally involves two types: various size and diverse aspect ratios of the text regions. In contrast to most existing approaches focusing on addressing one type of scale variance, this paper presents

Orthogonal moments enable computer-based systems to discriminate between similar objects. Mathematicians proved that the orthogonal polynomials of fractional-orders outperformed their corresponding counterparts in representing the fine details of a given function. In this work, novel orthogonal fractional-order Legendre-Fourier moments are proposed for pattern recognition applications. The basis functions

Recently, different studies have demonstrated the use of co-clustering, a data mining technique which simultaneously produces row-clusters of observations and column-clusters of features. The present work introduces a novel co-clustering model to easily summarize textual data in a document-term format. In addition to highlighting homogeneous co-clusters as other existing algorithms do we also distinguish

We consider the multi-view data completion problem, i.e., to complete a matrix U=[U1|U2] where the ranks of U, U1, and U2 are given. In particular, we investigate the fundamental conditions on the sampling pattern, i.e., locations of the sampled entries for finite completability of such a multi-view data given the corresponding rank constraints. We provide a geometric analysis on the manifold structure

Recently, graph convolutional networks (GCNs) have achieved state-of-the-art results for skeleton based action recognition by expanding convolutional neural networks (CNNs) to graphs. However, due to the lack of effective feature aggregation method, e.g. max pooling in CNN, existing GCN-based methods only learn local information among adjacent joints and are hard to obtain high-level interaction features

Biclustering can be defined as the simultaneous clustering of rows and columns in a data matrix and it has been recently applied to many scientific scenarios such as bioinformatics, text analysis and computer vision to name a few. In this paper we propose a novel biclustering approach, that is based on the concept of dominant-set clustering and extends such algorithm to the biclustering problem. In

Conventional minutia-based fingerprint recognition requires a complicated geometric matching and hard to be adopted in the bit-string based cancellable biometrics or bio-encryption, as the minutia data representing a fingerprint image is geometrical, unordered and variable in size. In this paper, we propose a new method to represent a fingerprint image by an ordered and fixed-length bit-string to cope

Chinese characters have a valuable property, this is, numerous Chinese characters are composed of a compact set of fundamental and structural radicals. This paper introduces a radical analysis network (RAN) that makes full use of this valuable property to implement radical-based Chinese character recognition. The proposed RAN employs an attention mechanism to extract radicals from Chinese characters

An u-shapelet is a sub-sequence of a time series used for the clustering of time series datasets. The purpose of this paper is to discover u-shapelets on uncertain time series. To achieve this goal, we propose a dissimilarity score called FOTS whose computation is based on the eigenvector decomposition and the comparison of the autocorrelation matrices of the time series. This score is robust to the

Head pose estimation plays a vital role in various applications, e.g., driver-assistance systems, human-computer interaction, virtual reality technology, and so on. We propose a novel geometry-based method for accurately estimating the head pose from a single 2D face image at a very low computational cost. Specifically, the rectangular coordinates of only four non-coplanar feature points from a predefined

Optimizing deep neural networks (DNNs) often suffers from the ill-conditioned problem. We observe that the scaling based weight space symmetry (SBWSS) in rectified nonlinear network will cause this negative effect. Therefore, we propose to constrain the incoming weights of each neuron to be unit-norm, which is formulated as an optimization problem over the Oblique manifold. A simple yet efficient method

End-to-end training of a deep CNN-Based model for salient object detection usually requires a huge number of training samples with pixel-level annotations, which are costly and time-consuming to obtain. In this paper, we propose an approach that can utilize large amounts of web data for learning a deep salient object detection model. With thousands of images collected from the Web, we first employ

A standard word embedding algorithm, such as ‘word2vec’, embeds each word as a dense vector of a preset dimensionality, the components of which are learned by maximizing the likelihood of predicting the context around it. However, as an inherent linguistic phenomenon, it is evident that there is a varying degree of difficulty in identifying words from their contexts. This suggests that a variable granularity

Image classifiers based on deep neural networks show severe vulnerability when facing adversarial examples crafted on purpose. Designing more effective and efficient adversarial attacks is attracting considerable interest due to its potential contribution to interpretability of deep learning and validation of neural networks’ robustness. However, current iterative attacks use a fixed step size for

The current deep learning based spatio-temporal action localization methods that using motion information (predominated is optical flow) obtain the state-of-the-art performance. However, since the optical flow is pre-computed, leading to these methods face two problems – the computational efficiency is low and the whole network is not end-to-end trainable. We propose a novel spatio-temporal action

The first attention model in the computer science community is proposed in 1998. In the following years, human attention has been intensively studied. However, these studies mainly refer human attention as the image regions that draw the attention of a human (outside the image) who is looking at the image. In this paper, we infer the attention of a human inside a third-person view video where the human

Low-rank matrix factorization (LRMF) has received much popularity owing to its successful applications in both computer vision and data mining. By assuming noise to come from a Gaussian, Laplace or mixture of Gaussian distributions, significant efforts have been made on optimizing the (weighted) L1 or L2-norm loss between an observed matrix and its bilinear factorization. However, the type of noise

Face alignment has gained great popularity in computer vision due to its wide-spread applications. In this paper, we propose a novel learning architecture, i.e., heterogenous output regression network (HORNet), for face alignment, which directly predicts facial landmarks from images. HORNet is based on kernel approximations and establishes a new compact multi-layer architecture. A nonlinear layer with

In this paper, we address the problem of depth recovery from a sequence of multi-focus images, known as shape-from-focus (SFF). The conventional SFF techniques typically exhibit poor performance over textureless regions, and it is difficult to preserve depth edges and fine details while maintaining spatial consistency. Therefore, we propose an SFF depth recovery framework composed of depth reconstruction

Image matching aims to find a similar area of the small image in the large image, which is one of the key steps in image fusion and vision-based navigation; however, most matching methods perform poorly when the images to be matched are blurred. Traditional approaches for blurred image matching usually follow a two-stage framework - first resorting to image deblurring and then performing image matching

Image and video cosegmentation is a newly emerging and rapidly progressing area, which aims at delineating common objects at pixel-level from a group of images or a set of videos. Plenty of related works have been published and implemented in varied applications, but there lacks a systematic survey on both image and video cosegmentation. This paper provides a comprehensive overview including the existing

Image corners have been widely used in various computer vision tasks. Current multi-scale analysis based corner detectors do not make full use of the multi-scale and multi-directional structural information. This degrades their detection accuracy and capability of refining corners. In this work, an improved shearlet transform with a flexible number of directions and a reasonable support is proposed

Beneficial from Fully Convolutional Neural Networks (FCNs), saliency detection methods have achieved promising results. However, it is still challenging to learn effective features for detecting salient objects in complicated scenarios, in which i) non-salient regions may have ”salient-like” appearance; ii) the salient objects may have different-looking regions. To handle these complex scenarios, we

the GrabCut can effectively extract the foreground according to features in a cartoon image; however, the performance is not so effective for a real image, because the feature extraction is independent of segmentation. To improve segmentation performance, this paper proposes an improved GrabCut which combines the segmentation and multiscale feature extraction into a unified model. In this model, the

To improve on the robustness of traditional machine learning approaches, emphasis has recently shifted to the integration of such methods with Deep Learning techniques. However, the classification problems, complexity and inconsistency in several spectral classifiers developed for hyperspectral images are some reasons warranting further research. This study investigates the application of Deep Support

We focus on an image classification task in which only several unlabeled images per class are available for learning and low computational complexity is required. We recall the state-of-the-art methods that are used to solve the task: autoencoder-based approaches and manifold-decomposition-based approaches. Next, we introduce our proposed method, which is based on a combination of the F-transform and

Skeleton-based action recognition is an increasing attentioned task that analyses spatial configuration and temporal dynamics of a human action from skeleton data, which has been widely applied in intelligent video surveillance and human-computer interaction. How to design an effective framework to learn discriminative spatial and temporal characteristics for skeleton-based action recognition is still

Tracking movement of insects in a social group (such as ants) is challenging, because the individuals are not only similar in appearance but also likely to perform intensive body contact and sudden movement adjustment (start/stop, direction changes). To address this challenge, we introduce an online multi-object tracking framework that combines both the motion and appearance information of ants. We

Recently, there have been various saliency detection methods proposed for still images based on deep learning techniques. However, the research on saliency detection for video sequences is still limited. In this study, we introduce a novel deep learning framework of saliency detection for video sequences, namely Deep Video Saliency Network (DevsNet). DevsNet mainly consists of two components: 3D Convolutional

With the breakthroughs in general action understanding, it has become an inevitable trend to analyze the actions in finer granularity. However, related researches have been largely hindered by the lack of fine-grained datasets and the difficulty of capturing subtle differences between fine-grained actions that are highly similar overall. In this paper, we address the above challenges by constructing

The binary neural network, largely saving the storage and computation, serves as a promising technique for deploying deep models on resource-limited devices. However, the binarization inevitably causes severe information loss, and even worse, its discontinuity brings difficulty to the optimization of the deep network. To address these issues, a variety of algorithms have been proposed, and achieved

Saliency detection has made remarkable progress along with the development of deep learning. While how to integrate the low-level intrinsic context with high-level semantic information to keep the object boundary sharp and restrain the background noise is still a challenging problem. Many attempts on network structures and refinement strategies have been explored, such as using Conditional Random Field