Multi-Label Classification (MLC) is an extension of the standard single-label classification where each data instance is associated with several labels simultaneously. MLC has gained much importance in recent years due to its wide range of application domains. However, the class imbalance problem has become an inherent characteristic of many multi-label datasets, where the samples and their corresponding

For signature verification systems, micro deformations can be defined as the small differences in the same strokes of signatures or special writing habits of different signers. These micro deformations can reveal the core distinction between the genuine signatures and skilled forgeries. In this paper, we prove that Convolutional Neural Networks (CNNs) have the potential to extract those micro deformations

Computer-aided diagnosis has been extensively investigated for more rapid and accurate screening during the outbreak of COVID-19 epidemic. However, the challenge remains to distinguish COVID-19 in the complex scenario of multi-type pneumonia classification and improve the overall diagnostic performance. In this paper, we propose a novel periphery-aware COVID-19 diagnosis approach with contrastive representation

Document binarization is a key step in most document analysis tasks. However, historical-document images usually suffer from various degradations, making this a very challenging processing stage. The performance of document image binarization has improved dramatically in recent years by the use of Convolutional Neural Networks (CNNs). In this paper, a dual-task, T-shaped neural network is proposed

Quadratic discriminant analysis (QDA) is a widely used statistical tool to classify observations from different multivariate Normal populations. The generalized quadratic discriminant analysis (GQDA) classification rule/classifier, which generalizes the QDA and the minimum Mahalanobis distance (MMD) classifiers to discriminate between populations with underlying elliptically symmetric distributions

In this paper1, we investigate cross-scene video foreground segmentation via supervised and unsupervised model communication. Traditional unsupervised background subtraction methods often face the challenging problem of updating the statistical background model online. In contrast, supervised foreground segmentation methods, such as those that are based on deep learning, rely on large amounts of training

Biclustering (co-clustering, two-mode clustering), as one of the classical unsupervised learning methods, has been applied in many different fields in recent years. Different types of biclustering methods have been developed such as probabilistic methods, two-way clustering methods, variance minimization methods, and so on. However, few regression-based methods have been proposed to the best of our

In this paper, we propose deep learning frameworks based on the randomized neural network. Inspired by the principles of Random Vector Functional Link (RVFL) network, we present a deep RVFL network (dRVFL) with stacked layers. The parameters of the hidden layers of the dRVFL are randomly generated within a suitable range and kept fixed while the output weights are computed using the closed-form solution

Many warping methods for image stitching have been proposed to construct panoramic image mosaics free of artifacts. Existing methods heavily rely on coordinate correspondences between keypoints in stitching, which may not provide adequate constraints for alignment. In this paper, we discover and employ a new constraint — angle correspondences to address the above problem. The angle of a feature point

This paper presents an end-to-end neural network system to identify writers through handwritten word images, which jointly integrates global-context information and a sequence of local fragment-based features. The global-context information is extracted from the tail of the neural network by a global average pooling step. The sequence of local and fragment-based features is extracted from a low-level

A reliable tracker has the ability to adapt to change of objects over time, and is robust and accurate. We build such a tracker by extracting semantic features using robust Siamese networks and multi-granularity color features. It incorporates a semantic model that can capture high quality semantic features and an appearance model that can describe object at pixel, local and global levels effectively

Spiking neural network (SNN) is a type of artificial neural network that uses biologically inspired neuron models and learning rules to develop artificial intelligence with capability parallel to human brain. Deep neural networks (DNNs), on the other hand, uses less biologically plausible neurons and training methods such as gradient descent, and has shown good accuracy in computer vision tasks. However

The early detection of COVID-19 is a challenging task due to its deadly spreading nature and existing fear in minds of people. Speech-based detection can be one of the safest tools for this purpose as the voice of the suspected can be easily recorded. The Mel Frequency Cepstral Coefficient (MFCC) analysis of speech signal is one of the oldest but potential analysis tools. The performance of this analysis

Occlusion due to hair in dermoscopic images affects the diagnostic operation and the accuracy of its analysis of a skin lesion. Also, dermis hair has the following different characteristics: thin; overlapping; faded; of similar contrast or colour to the underlying skin; and obscuring/covering textured lesions. These make digital hair removal (DHR), which involves hair segmentation and hair gap inpainting

Clustering aims to partition an input dataset into distinct groups according to some distance or similarity measurements. One of the most widely used clustering method nowadays is the k-means algorithm because of its simplicity and efficiency. In the last few decades, k-means and its various extensions have been formulated to solve the practical clustering problems. However, existing clustering methods

Typical deep-neural-network (DNN) based generative image models often (i) show limited ability to learn a disentangled latent representation, (ii) show limited controllability leading to undesirable side effects when manipulating selected attributes during image generation, and (iii) require large attribute-annotated training sets. We propose a generative DNN model for face images by explicitly disentangling

Though image segmentation models are plentiful and have many applications nowadays, it can be difficult to segment images with complex boundaries and serious intensity inhomogeneity. To some extent, the region-scalable fitting energy model can segment images suffering from intensity inhomogeneity since it considers image intensity as a function, but it relies on initial conditions dramatically. Nowadays

Hyperspectral anomaly detection (HAD) is a research endeavor of high practical relevance within remote sensing scene interpretation. In this work, we propose an unsupervised approach, dual feature extraction network (DFEN) for HAD, to gradually build up ever-greater discrimination between the original data and background. In particular, we impose an end-to-end discriminative learning loss on two networks

This paper presents a new nested U-shape attention network (NUA-Net) with improved robustness of lesions for effective vascular segmentation in retinal imaging. Unlike most of the current deep learning approaches which rely on vanilla upsample module to recover distinguishable features for segmentation, our attention-based multi-scale network extends the U-shape segmentation network by introducing

Object detection is an important field in computer vision. Nevertheless, a research area that has so far not received much attention is the study into the effectiveness of anchor matching strategy and imbalance in anchor-based object detection, in particular small object detection. It is clear that the objects with larger sizes tend to match more anchors than smaller ones. This matching imbalance may

Matching local features on two or more images is fundamental for many applications in the field of computer vision and pattern recognition. Identifying and rejecting mismatches is an important part in the framework of feature matching, due to the putative correspondences always contaminated by mismatches with the error-prone local feature detectors. In this paper, we introduce a novel method, namely

Sparse representation and cooperative learning are two representative technologies in the field of multi-view spectral clustering. The former can effectively extract features of multiple views by the removal of redundant information contained in each view. The latter can incorporate the diversity of each view. However, traditional sparse representation and cooperative learning algorithms are inadequate

One of the important problems in computer-aided diagnosis of connective tissue disease is automatic recognition of staining patterns present in HEp-2 cells. In this regard, the paper introduces a novel approach for the recognition of staining patterns by HEp-2 cell indirect immunofluorescence image analysis. The proposed method assumes that a fixed set of local texture descriptors or scales may not

One-class classification is a challenging subfield of machine learning in which so-called data descriptors are used to predict membership of a class based solely on positive examples of that class, and no counter-examples. A number of data descriptors that have been shown to perform well in previous studies of one-class classification, like the Support Vector Machine (SVM), require setting one or more

In recent years, a large number of meta-learning methods have been proposed to address few-shot learning problems and have shown superior performance. However, the explicit prior knowledge (e.g., concept graph) and weakly-supervised information are rarely explored in existing methods, which are usually free or cheap to collect. In this paper, we introduce a concept graph for the weakly-supervised few-shot

We propose a random Fourier sampling scheme to enhance the accuracy of the high frequency pattern estimation for image reconstruction. This method is designed to work in a constrained ℓ1 minimization based on the Fourier-Haar interplay revealing a column-wise maximum coherent structure that we provide. Essential in the scheme is to generate a data-driven density function by a small percentage of Fourier

Feature selection is an important procedure in machine learning because it can reduce the complexity of the final learning model and simplify the interpretation. In this paper, we propose a novel non-linear feature selection method that targets multi-class classification problems in the framework of support vector machines. The proposed method is achieved using a kernelized multi-class support vector

Hausdorff distance (HD) is commonly used as a similarity metric in image or 3D volume comparison. Although HD is accurate and popular for a variety of tasks, its main weakness is the consumption of processing power. In this work, a novel, parallel and locality-oriented Hausdorff distance algorithm is proposed. Novel as it is the first time in the literature that an algorithmic implementation using

Spectral clustering (SC) is a popular approach for gaining insights from complex networks. Conventional SC focuses on second-order structures (e.g. edges) without direct consideration of higher-order structures (e.g. triangles). This has motivated SC extensions that directly consider higher-order structures. However, both approaches are limited to considering a single order. To address this issue,

The problem of inhomogeneous cluster densities has been a long-standing issue for distance-based and density-based algorithms in clustering and anomaly detection. These algorithms implicitly assume that all clusters have approximately the same density. As a result, they often exhibit a bias towards dense clusters in the presence of sparse clusters. Many remedies have been suggested; yet, we show that

Attention-based scene text recognizers have gained huge success, which leverages a more compact intermediate representation to learn 1d- or 2d- attention by a RNN-based encoder-decoder architecture. However, such methods suffer from attention-driftproblem because high similarity among encoded features leads to attention confusion under the RNN-based local attention mechanism. Moreover, RNN-based methods

Robust and accurate multi-source matching is a difficult task due to significant nonlinear radiometric differences, background clutter, and geometric deformation in corresponding regions. Motivated by these existing problems, a discriminating yet robust combined descriptor for multi-source image matching, called deformed contour segment similarity (DCSS), is proposed in this work. First, the proposed

Accurately matching visual and textual data in cross-modal retrieval has been widely studied in the multimedia community. To address these challenges posited by the heterogeneity gap and the semantic gap, we propose integrating Shannon information theory and adversarial learning. In terms of the heterogeneity gap, we integrate modality classification and information entropy maximization adversarially

Segmenting optic disc (OD) in abnormal fundus images is a challenge task because of many distractions such as illumination variations, blurry boundary, occlusion of retinal vessels and big bright lesions. Data-driven deep learning is effective and robust to illumination variations, blurry boundary and occlusion in the normal fundus images but sensitive to big bright lesions in abnormal images. In this

Visual Question Answering (VQA) as an important task in understanding vision and language has been proposed and aroused wide interests. In previous VQA methods, Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) are generally used to extract visual and textual features respectively, and then the correlation between these two features is explored to infer the answer. However, CNN

Weakly-supervised semantic segmentation, as a promising solution to alleviate the burden of collecting per-pixel annotations, aims to train a segmentation model from partial weak annotations. Scribble on the object is one of the commonly used weak annotations and has shown to be sufficient for learning a decent segmentation model. Despite being effective, scribble-based weakly-supervised learning methods

With the advantages of low storage cost and extremely fast retrieval speed, deep hashing methods have attracted much attention for image retrieval recently. However, large-scale face image retrieval with significant intra-class variations is still challenging. Neither existing pairwise/triplet labels-based nor softmax classification loss-based deep hashing works can generate compact and discriminative

Low-level features and deep features each have their own advantages and disadvantages in image representation. However, combining their advantages within a CBIR framework remains challenging. To address this problem, we propose a novel image-retrieval method: the deep-seated features histogram (DSFH). Its main highlights are: 1) Low-level features are extracted by simulating the human orientation selection

The popularity of smartphones with digital cameras makes photographing using smartphones an important daily activity. Moiré patterns can easily appear when shooting objects with rich textures, such as computer screens, and will severely degrade the image quality. Image demoiréing is an important image restoration task that aims to remove moiré patterns and reveal the underlying clean image. Two key

In recent years, impressive advances have been made in single-image super-resolution. Deep learning is behind much of this success. Deep(er) architecture design and external prior modeling are the key ingredients. The internal contents of the low-resolution input image are neglected with deep modeling, despite earlier works that show the power of using such internal priors. In this paper, we propose

Optical coded targets allow to determine the relative pose of a camera, on a metric scale, from one image only. Furthermore, they are easily and efficiently detected, opening to a wide range of applications in robotics and computer vision. In this work we describe the effect of pixel saturation and non-ideal lens Point Spread Function, causing the apparent position of the corners and the edges of the

Semantic segmentation tasks based on weakly supervised conditions have been put forward to achieve a lightweight labeling process. For simple images that only include a few categories, research based on image-level annotations has achieved acceptable performance. However, when facing complex scenes, since image contains a large number of classes, it becomes challenging to learn visual appearance based

Many video understanding tasks work in the offline setting by assuming that the input video is given from the start to the end. However, many real-world problems require the online setting, making a decision immediately using only the current and the past frames of videos such as in autonomous driving and surveillance systems. In this paper, we present a novel solution for online action detection by

Most Zero-Shot Action Recognition (ZSAR) methods establish visual-semantic joint embedding space, which is based on commonly used visual features and semantic embeddings, to learn the correlation between actions. Nevertheless, extracting visual features without structural guidance would lead to sparse video features, which reflect the correlation of actions, fall into oblivion. Based on the Ventral

Correlation clustering (CC) is a clustering method using a signed graph as input without specifying the number of clusters a priori. It had been widely used in real applications, such as social network and text mining. However, its exact optimization or approximate algorithms often give unsatisfactory results, especially for large-scale signed graphs. This paper tackles this problem and proposes a

Lung cancer is among the most common and deadliest cancers with a low 5-year survival rate. Timely diagnosis of lung cancer is, therefore, of paramount importance as it can save countless lives. In this regard, Computed Tomography (CT) scan is widely used for early detection of lung cancer, where human judgment is currently considered as the gold standard approach. Recently, there has been a surge

For a Level 3 automated vehicle, according to the SAE International Automation Levels definition (J3016), the identification of non-driving activities (NDAs) that the driver is engaging with is of great importance in the design of an intelligent take-over interface. Much of the existing literature focuses on the driver take-over strategy with associated Human-Machine Interaction design. This paper

As advanced facial manipulation technologies develop rapidly, one can easily modify an image by changing the identity or the facial expression of the target person, which threatens social security. To address this problem, face forgery detection becomes an important and challenging task. In this paper, we propose a novel network, called Pixel-Region Relation Network (PRRNet), to capture pixel-wise

VLAD (Vector of Locally Aggregated Descriptors) has been widely adopted in image representation. However, the VLAD algorithm seeks for the algebraic sum of the residue vectors between the descriptors and the centroid of cluster they belong to, and this could decrease the discriminative power of feature representations. To this end, this paper originally proposes a VLAAD (Vector of Locally and Adaptively

Prototype-based few-shot learning methods are promising in that they are simple yet effective to handle any-shot problems, and many prototype associated works are raised since then. However, these traditional prototype-based methods generally use only one single prototype to represent a class, which essentially cannot effectively estimate the complicated distribution of a class. To tackle this problem

Deep learning techniques have been increasingly applied to the diagnosis of Alzheimer’s disease (AD) and the conversion from mild cognitive impairment (MCI) to AD. Despite their prevalence, existing methods usually suffer from using either irrelevant brain regions or less-accurate landmarks. In this paper, we propose the iterative sparse and deep learning (ISDL) model for joint deep feature extraction

Fine-grained visual categorization (FGVC) has attracted extensive attention in recent years. The general pipeline of current FGVC techniques is to 1) locate the discriminative regions; 2) extract features from each region independently; and 3) feed the integrated features to a classifier. In this paper, we re-investigate the pipeline from the view of human visual recognition mechanisms. The perceiving

Weakly supervised object localization (WSOL) methods utilize the internal feature responses of a classifier trained only on image-level labels. Classifiers tend to focus on the most discriminative part of the target object, instead of considering its full extent. Adversarial erasing (AE) techniques have been proposed to ameliorate this problem. These techniques erase the most discriminative part during

As a new method of human-computer interaction, inertial sensor-based in-air handwriting can provide natural and unconstrained interaction to express more complex and rich information in 3D space. However, most of the existing literature is mainly focused on in-air handwriting recognition (IAHR), which makes these works suffer from the poor readability of inertial signals and the lack of labeled samples

Motion information used in the existed video action recognition schemes is mixing of global motion(GM) and local motion(LM). In fact, GM & LM have their respective semantic concepts. Thus, it is promising to decouple GM and LM from the mixed motions. Numerous efforts have been made on the design of global motion models for video encoding, video dejittering, video denoising, and so on. Nevertheless

This manuscript presents a new method for fitting ellipses to two-dimensional data using the confocal hyperbola approximation to the geometric distance of points to ellipses. The proposed method was evaluated and compared to established methods on simulated and real-world datasets. First, it was revealed that the confocal hyperbola distance considerably outperforms other distance approximations such

Meta-learning is an effective tool to address the few-shot learning problem, which requires new data to be classified considering only a few training examples. However, when used for classification, it requires large labeled datasets, which are not always available in practice. In this paper, we propose an unsupervised meta-learning algorithm that learns from an unlabeled dataset and adapts to downstream