Boguslaw Cyganek

AGH University of Science and Technology in Kraków | AGH · Department of Electronics

In this paper, a custom underwater 2D sonar signal acquisition, enhancement, and tracking system are presented. The system is based on a modern 2D multibeam sonar by Teledyne allowing data gathering and real-time operation. First, our custom setup allows 2D sonar signal acquisition for gathering underwater datasets, which are rare in the public dom...

Cancer diseases constitute one of the most significant societal challenges. In this paper we introduce a novel histopathological dataset for prostate cancer detection. The proposed dataset, consisting of over 2.6 million tissue patches extracted from 430 fully annotated scans, 4675 scans with assigned binary diagnosis, and 46 scans with diagnosis g...

Today’s deep learning architectures, if trained with proper dataset, can be used for object detection in marine search and rescue operations. In this paper a dataset for maritime search and rescue purposes is proposed. It contains aerial-drone videos with 40,000 hand-annotated persons and objects floating in the water, many of small size, which mak...

In recent years many methods have been proposed for eye detection. In some cases however, such as driver drowsiness detection, lighting conditions are so challenging that only the thermal imaging is a robust alternative to the visible light sensors. However, thermal images suffer from poor contrast and high noise, which arise due to the physical pr...

In this paper the video change detection method that allows for data privacy protection is proposed. Signal change detection is based on the tensor models constructed in the orthogonal tensor subspaces. Tensor methods allow for processing of any kind of multi-dimensional signals since computation of special features is not required. The proposed si...

Histograms of oriented gradients (HOG) are still one of the most frequently used low-level features for pattern recognition in images. Despite their great popularity and simple implementation performance of the HOG features almost always has been measured on relatively high quality data which are far from real conditions. To fill this gap we experi...

While most existing image recognition benchmarks consist of relatively high quality data, in the practical applications images can be affected by various types of distortions. In this paper we experimentally evaluate the extent to which image distortions affect classification based on HOG feature descriptors. In an experimental study based on sever...

An overview of modern tensor based methods for multi-dimensional signal processing is presented. Special focus is laid on recent achievements in signal change detection, as well as on efficient methods of their compression based on various tensor decompositions. Apart from theory, applications as well as implementation issues are presented as well.

In this work, we propose an algorithm for training deep neural networks for classification of breast cancer in histopathological images affected by data unbalance with support of active learning. The output of the neural network on unlabeled samples is used to calculate weighted information entropy. It is utilized as uncertainty score for automatic...

In this work, we propose an algorithm for training deep neural networks for classification of breast cancer in histopathological images affected by data unbalance with support of active learning. The output of the neural network on unlabeled samples is used to calculate weighted information entropy. It is utilized as uncertainty score for automatic...

In this paper an efficient method for signal change detection in multidimensional data streams is proposed. A novel tensor model is suggested for input signal representation and analysis. The model is built from a part of the multidimensional stream by construction of the representing orthogonal tensor subspaces, computed with the higher-order sing...

In this work, we propose an algorithm for training deep neural networks for classification of breast cancer in histopathological images affected by data unbalance with support of active learning. The output of the neural network on unlabeled samples is used to calculate weighted information entropy. It is utilized as uncertainty score for automatic...

Real signals are usually contaminated with various types of noise. This phenomenon has a negative impact on the operation of systems that rely on signals processing. In this paper, we propose a tensor-based method for speckle noise reduction in the side-scan sonar images. The method is based on the Tucker decomposition with automatically determined...

Due to variety of modern real-life tasks, where analyzed data is often not a static set, the data stream mining gained a substantial focus of machine learning community. Main property of such systems is the large amount of data arriving in a sequential manner, which creates an endless stream of objects. Taking into consideration the limited resourc...

The paper presents a novel approach to driver's fatigue recognition based on yawn detection using thermal imaging. This work writes into the active drivers’ assisting systems which can warn on driver's drowsiness based on continuous observations. The method can operate in day and night conditions without distracting a driver due to usage of thermal...

In this paper we experimentally evaluated the impact of data imbalance on the convolutional neural networks performance in the histopathological image recognition task. We conducted our analysis on the Breast Cancer Histopathological Database. We considered four phenomena associated with data imbalance: how does it affect classification performance...

Due to the advances made in recent years, methods based on deep neural networks have been able to achieve a state-of-the-art performance in various computer vision problems. In some tasks, such as image recognition, neural-based approaches have even been able to surpass human performance. However, the benchmarks on which neural networks achieve the...

Underwater images suffer from various degradation factors, such as blur, haze, color degradation, and marine snow. Marine snow is a type of noise, caused mostly by biological particles that fall into the ocean bottom, and which impedes proper object detection in underwater vision. A method for real-time marine snow removal from underwater color and...

The paper presents a method for change detection in multidimensional streams of data based on a tensor model constructed from the Higher-Order Singular Value Decomposition of raw data tensors. The method was applied to the problem of video shot detection showing good accuracy and high speed of execution compared with other more time demanding tenso...

In the paper a tensor based method for video stream clustering and compression is presented. The method does video partitioning in temporal domain based on its content. Such coherent video partitions are amenable for better compression. The proposed method detects shot boundaries building a tensor model from a number of frames in the stream. To bui...

Data classification in presence of noise can lead to much worse results than expected for pure patterns. In this paper we investigate this problem in the case of deep convolutional neural networks in order to propose solutions that can mitigate influence of noise. The main contributions presented in this paper are experimental examination of influe...

This paper presents a method for content change detection in multidimensional video signals. Video frames are represented as tensors of order consistent with signal dimensions. The method operates on unprocessed signals and no special feature extraction is assumed. The dynamic tensor analysis method is used to build a tensor model from the stream....

In the paper an original method of the oxyacetylene welding measurement and control is presented. The method is based on the computer processing of the flame images of the oxyacetylene torch. In this paper flame analysis is presented which is based on adaptive thresholding, statistical shape parameters computations, as well as color analysis of cha...

We investigate performance of the classical PCA based background subtraction procedure and compare it with the robust PCA versions which are computationally demanding. We show that the simple PCA based version endowed with the fast eigen-decomposition method allows real-time operation on VGA video streams while offering accuracy comparable with som...

The paper presents an efficient method for real-time image analysis for manoeuvring of the underwater robot. Image analysis is done after computing the structural tensor components which unveil rich texture and texture-less areas. To allow a power efficient underwater operation in real-time the method is implemented on the Jetson TK1 self-standing...

This paper address the data mining task of classifying data stream with concept drift. The proposed algorithm, named Concept-adapting Evolutionary Algorithm For Decision Tree (CEVOT) does not require any knowledge of the environment in which it operates (e.g. numbers and rates of drifts). The novelty of the approach is combining tree learner and ev...

In this paper new methods for fast computation of the chordal kernels are proposed. Two versions of the chordal kernels for tensor data are discussed. These are based on different projectors of the flattened matrices obtained from the input tensors. A direct transformation of multidimensional objects into the kernel feature space leads to better da...

Contemporary classification systems have to make a decision not only on the basis of the static data, but on the data in motion as well. Objects being recognized may arrive continuously to a classifier in the form of data stream. Usually, we would like to start exploitation of the classifier as soon as possible, the models which can improve their m...

Presence of noise poses a common problem in image recognition tasks. In this paper we propose and analyse architecture of convolutional neural network capable of image denoising. We evaluate its performance with various types of artificial distortions present, with both known and unknown noise conditions. Finally, we measure how including denoising...

For the contemporary enterprises, possibility of appropriate business decision making on the basis of the knowledge hidden in stored data is the critical success factor. Therefore, the decision support software should take into consideration that data usually comes continuously in the form of so-called data stream, but most of the traditional data...

you can download this paper using the following link (50 downloads are available) http://www.tandfonline.com/eprint/3SVz9xibkdcA3zv9eJnf/full The Electronic Health Record (EHR) groups all digital documents related to a given patient such as anamnesis, results of the laboratory tests, prescriptions, recorded medical signals as ECG or images, etc. D...

you can download this article using the following link (50 downloads are available) http://www.tandfonline.com/eprint/QeQ9EmvJVsVJQGfRZKYg/full Editorial to the special issue of the Applied Artificial Intelligence on Intelligent Methods Applied to Health-Care Information Systems

The big data is characterized by 4Vs (volume, velocity, variety, and variability). In this paper we focus on the velocity, but actually it usually comes together with volume. It means, that the crucial problem of the contemporary data analytics is to answer the question how to discover useful knowledge from fast incoming data. The paper presents an...

In this paper we discuss algorithmically efficient methods of multidimensional patter recognition in kernel tensor subspaces. The kernel principal component analysis, which originally operates only on vector data, is joined with the tensor chordal kernel which opens a way of direct usage of the multidimensional signals, such as color video streams,...

Classification of distorted patterns poses real problem for majority of classifiers. In this paper we analyse robustness of deep neural network in classification of such patterns. Using specific convolutional network architecture, an impact of different types of noise on classification accuracy is evaluated. For highly distorted patterns to improve...

In many practical situations visual pattern recognition is vastly burdened by low quality of input images due to noise, geometrical distortions, as well as low quality of the acquisition hardware. However, although there are techniques of image quality improvements, such as nonlinear filtering, there are only few attempts reported in the literature...

In this paper we address the problem of the reduction of multiplicative noise in digital images. This kind of image distortion, also known as speckle noise, severely decreases the quality of medical ultrasound images and therefore their effective enhancement and restoration is of vital importance for proper visual inspection and quantitative measur...

This work reports the research on active learning approach applied to the data stream classification. The chosen characteristics of the proposed frameworks were evaluated on the basis of the wide range of computer experiments carried out on the three benchmark data streams. Obtained results confirmed the usability of proposed method to the data str...

The paper presents a modified method of building ensembles of tensor classifiers for direct multidimensional pattern recognition in tensor subspaces. The novelty of the proposed solution is a method of lowering tensor subspace dimensions by rotation of the training pattern to their optimal directions. These are obtained computing and analyzing phas...

This paper presents a visual system for real-time eye detection and tracking in the near-infrared (NIR) video streams for drivers’ monitoring. The system starts with crude eye position estimation based on an eye model suitable for NIR processing. In the next step, eye regions are verified with the classifier operating in the higher-order decomposit...

The big data is characterized by 4Vs (volume, velocity, variety, andvariability). In this paper we focus on the velocity, but actually it usually comes together with volume. It means, that the crucial problem of the contemporary data analytics is to answer the question how to discover useful knowledge from fast incoming data. The paper presents an...

In contemporary machine learning multidimensional rather than pure vector like data are frequently encountered. Traditionally, such multidimensional objects, such as color images or video sequences, are first transformed to a vector representation, and then processed by the classical learning algorithms operating with vectors. However, such multi-t...

One-class classification belongs to the one of the novel and very promising topics in contemporary machine learning. In recent years ensemble approaches have gained significant attention due to increasing robustness to unknown outliers and reducing the complexity of the learning process. In our previous works, we proposed a highly efficient one-cla...

In this paper a novel Tensor-Based Image Segmentation Algorithm (TBISA) is presented, which is dedicated for segmentation of colour images. A purpose of TBISA is to distinguish specific objects based on their characteristics, i.e. shape, colour, texture, or a mixture of these features. All of those information are available in colour channel data....

The paper presents a vehicle logo recognition system based on novel combination of tensor based feature extraction and ensemble of tensor subspace classifiers. Each originally two-dimensional vehicle logotype is transformed to a three-dimensional feature tensor applying the extended structural tensor method. All such exemplary logo-tensors which co...

In this paper a framework for visual patterns recognition of higher dimensionality is discussed. In the training stage, the input prototype patterns are used to construct a multidimensional array—a tensor—whose each dimension corresponds to a different dimension of the input data. This tensor is then decomposed into a lower-dimensional subspace bas...

The paper presents a method for data classification with ensemble of one-class classifiers based on data segmentation. Each data class is partitioned with the nonnegative matrix factorization (NMF) algorithm with sparse constraints. It allows splitting of the input data into compact and consistent data clusters with automatic determination of a num...

In the paper a novel filtering design based on the concept of exploration of the pixel neighborhood by digital paths is presented. The paths start from the boundary of a filtering window and reach its center. The cost of transitions between adjacent pixels is defined in the hybrid spatial-color space. Then, an optimal path of minimum total cost, le...

In this paper a system for real-time recognition of objects in multidimensional video signals is proposed. Object recognition is done by pattern projection into the tensor subspaces obtained from the factorization of the signal tensors representing the input signal. However, instead of taking only the intensity signal the novelty of this paper is f...

This paper introduces a novel method for forming efficient one-class classifier ensembles. A common problem in one-class classification is a complex structure of the target class, which often leads to creation of a too expanded decision boundary. We propose to employ a clustering step in order to partition the target class into atomic subsets and u...

In the paper we present a novel method of grayscale image colorization utilizing the idea of Lipschitz cover applied to a digital image. Our algorithm represents the interactive type of colorization in which the user indicates the hints in form of scribbles of a given color. The method colorizes the grayscale image by analyzing the intensity dis- t...

In this paper, we propose a new ensemble for an effective segmentation of hyperspectral images. It uses one-class classifiers as base learners. We prove, that despite the multi-class nature of hyperspectral images using one-class approach can be beneficial. One need simply to decompose a multi-class set into a number of simpler one-class tasks. One...

Abstract In the paper an extension of object recognition based on the Higher-Order Singular Value Decomposition (HOSVD) to the 4th dimension is discussed. HOSVD based object recognition expands the concept of object recognition in the pattern spaces spanned by the PCA decomposition of vector patterns into the higher dimensions. However, contrary to...

The paper deals with the problem of data stream classification. In the previous works we proposed the WAE (Weighted Aging Ensemble) algorithm which may change the line-up of the classifier committee dynamically according to coming of new individual classifiers. The ensemble pruning method uses the diversity measure called the Generalized Diversity...

The paper presents a system for vehicle logo recognition from real digital images. The process starts with license plates localization, followed by vehicle logo detection. For this purpose the structural tensor is employed which allows fast and reliable detections even in low quality images. Detected logo areas are classified to the car brands with...

This paper presents a hybrid visual system for monitoring driver's states of fatigue, sleepiness and inattention based on driver's eye recognition. Safe operation in car conditions and processing in daily and night conditions are obtained thanks to the custom setup of two cameras operating in the visible and near infra-red spectra, respectively. In...

The paper presents a hybrid ensemble of diverse classifiers for logo and trademark symbols recognition. The proposed ensemble is composed of four types of different member classifiers. The first one compares color distribution of the logo patterns and is responsible for sifting out images of different color distribution. The second of the classifie...

In this paper a novel parallel algorithm for the tensor based classifiers for object recognition in digital images is presented. Classification is performed with an ensemble of base classifiers, each operating in the orthogonal subspaces obtained with the Higher-Order Singular Value Decomposition (HOSVD) of the prototype pattern tensors. Parallelis...

The paper presents a work-in-progress on a classification system for object detection in vision based industrial applications. The main idea of the presented system is application of the ensemble of one-class classifiers trained with specific features of objects of the whole scene. During recognition stage the ensemble tries to recognize the traine...

The paper presents a system for driver's eye recognition from near infrared (NIR) images. The system is organized in a cascade of two classification modules. The first one is responsible for initial eye detection and the second one for eye validation. Detection is based on a novel eye model suited for the NIR images. This process is augmented by th...

This chapter focuses on different computer vision techniques, which make use of tensors, as well as their decomposition and analysis. It presents a variety of tensor methods in application to computer vision (CV) and pattern recognition (PR) tasks. The chapter starts with definitions of tensors, first viewed as mathematical objects, especially stre...

This chapter contains short discussions on selected problems which can be useful for understanding or applying the methods and techniques discussed in the book Object Detection and Recognition in Digital Images: Theory and Practice . It first presents the morphological scale-space which shows some nice properties and in some applications can be mor...

This chapter is devoted to selected problems in object detection and tracking. Objects are characterized by their salient features, such as color, shape, texture, or other traits. Detection can be viewed as a classification problem in which the task is to tell the presence or absence of a specific object in an image. Classification within a group o...

This chapter discusses various object recognition methods with examples and practical realizations, mostly oriented toward automotive systems. It starts with the method of histograms of oriented structures coming from the structural tensor. Its connection with the morphological scale-space makes it more robust. The chapter discusses invariant based...

This chapter presents a number of important classification techniques and algorithms. It starts with formulation of the probabilistic classification framework. and then discusses the group of subspace classification methods. One of the most important methods from this category is the principal component analysis (PCA) and both theoretical derivatio...

The paper addresses the issue of searching for similar images and objects ina repository of information. The contained images are annotated with the helpof the sparse descriptors. In the presented research, different color and edgehistogram descriptors were used. To measure similarities among images, variouscolor descriptors are compared. For this p...

In this paper we propose a system for visual object detection and tracking based on the extended structural tensor and the ensemble of one-class support vector machines. First, the input color image is transformed with the anisotropic process into the extended structural tensor. Then the tensor space is clustered into the number of partitions which...

The paper presents architecture and properties of the ensemble of the classifiers operating in the tensor orthogonal spaces obtained with the Higher-Order Singular Value Decomposition of prototype tensors. In this paper two modifications to this architecture are proposed. The first one consists in embedding of the Extended Euclidean Distance metric...