Wyniki wyszukiwania dla: ai safety · robustness · uncertainty estimation · pedestrian detection · object detection. - MOST Wiedzy

Wyszukiwarka

Wyniki wyszukiwania dla: ai safety · robustness · uncertainty estimation · pedestrian detection · object detection.

wyświetlamy 1000 najlepszych wyników Pomoc

Wyniki wyszukiwania dla: ai safety · robustness · uncertainty estimation · pedestrian detection · object detection.

  • Toward Robust Pedestrian Detection With Data Augmentation

    Publikacja

    In this article, the problem of creating a safe pedestrian detection model that can operate in the real world is tackled. While recent advances have led to significantly improved detection accuracy on various benchmarks, existing deep learning models are vulnerable to invisible to the human eye changes in the input image which raises concerns about its safety. A popular and simple technique for improving robustness is using data...

    Pełny tekst do pobrania w portalu

  • Zdzisław Kowalczuk prof. dr hab. inż.

    W 1978 ukończył studia w zakresie automatyki i informatyki na Wydziale Elektroniki Politechniki Gdańskiej, następnie rozpoczął pracę na macierzystej uczelni. W 1986 obronił pracę doktorską, w 1993 habilitował się na Politechnice Śląskiej na podstawie pracy Dyskretne modele w projektowaniu układów sterowania. W 1996 mianowany profesorem nadzwyczajnym, w 2003 otrzymał tytuł profesora nauk technicznych. W 2006 założył i od tego czasu...

  • Robustness in Compressed Neural Networks for Object Detection

    Publikacja

    Model compression techniques allow to significantly reduce the computational cost associated with data processing by deep neural networks with only a minor decrease in average accuracy. Simultaneously, reducing the model size may have a large effect on noisy cases or objects belonging to less frequent classes. It is a crucial problem from the perspective of the models' safety, especially for object detection in the autonomous driving...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Evaluating calibration and robustness of pedestrian detectors

    Publikacja

    In this work robustness and calibration of modern pedestrian detectors are evaluated. Pedestrian detection is a crucial perception com- ponent in autonomous driving and here we study its performance under different image corruptions. Furthermore, we provide analysis of classifi- cation calibration of pedestrian detectors and we show a positive effect of using style-transfer augmentation technique. Our analysis is aimed as a step...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Cascade Object Detection and Remote Sensing Object Detection Method Based on Trainable Activation Function

    Publikacja
    • S. N. Shivappriya
    • M. J. P. Priyadarsini
    • A. Stateczny
    • C. Puttamadappa
    • B. D. Parameshachari

    - Remote Sensing - Rok 2021

    Object detection is an important process in surveillance system to locate objects and it is considered as major application in computer vision. The Convolution Neural Network (CNN) based models have been developed by many researchers for object detection to achieve higher performance. However, existing models have some limitations such as overfitting problem and lower efficiency in small object detection. Object detection in remote...

    Pełny tekst do pobrania w portalu

  • The circle object detection with the use of Msplit estimation

    Publikacja

    - Rok 2018

    The paper presents the use of Msplit(q) - estimation in the filtration and aggregation of point clouds containing a known number of elliptical shapes with preliminary unknown - locations and dimensions. These theoretical solutions may have practical relevance especially in the modelling of terrestrial laser scanning data of objects that have similar shape to circles. Mentioned shapes can be scanned of tree trunks, columns, gutters,...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Robust Object Detection with Multi-input Multi-output Faster R-CNN

    Publikacja

    Recent years have seen impressive progress in visual recognition on many benchmarks, however, generalization to the out-of-distribution setting remains a significant challenge. A state-of-the-art method for robust visual recognition is model ensembling. However, recently it was shown that similarly competitive results could be achieved with a much smaller cost, by using multi-input multi-output architecture (MIMO). In this work,...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Robust Object Detection with Multi-input Multi-output Faster R-CNN

    Publikacja

    Recent years have seen impressive progress in visual recognition on many benchmarks, however, generalization to the out-of-distribution setting remains a significant challenge. A state-of-the-art method for robust visual recognition is model ensembling. However, recently it was shown that similarly competitive results could be achieved with a much smaller cost, by using multi-input multi-output architecture (MIMO). In this work,...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Akustyczna analiza natężenia ruchu drogowego dla systemów zarządzania ruchem

    Publikacja

    - Rok 2019

    W pracy przybliżono wybrane zagadnienia z dziedziny zarządzania transportem drogowym w Polsce i na świecie. W tym kontekście pzredstawiono potrzeby rynkowe, wymagania jak i możliwości w zakresie pozyskiwania informacji o aktualnym stanie sieci drogowych. Zaproponowano akustyczną metodę nadzorowania ruchu drogowego i jej możliwości w kontekście systemów zarządzania ruchem. Przedstawiono schemat akwizycji sygnału wraz z danymi odniesienia....

  • An Analysis of Uncertainty and Robustness of Waterjet Machine Positioning Vision System

    Publikacja

    The paper presents a new Automatic Waterjet Positioning Vision System (AWPVS) and investigates components of workpiece positioning accuracy. The main purpose of AWPVS is to precisely identify the position and rotation of a workpiece placed on a waterjet machine table. Two webcams form a basis for the system, and constitute its characteristics. The proposed algorithm comprises various image processing techniques to assure a required...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Resolving conflicts in object tracking for automatic detection of events in video

    An algorithm for resolving conflicts in tracking of moving objects is presented. The proposed approach utilizes predicted states calculated by Kalman filters for estimation of trackers position, then it uses color and texture descriptors in order to match moving objects with trackers. Problematic situations, such as splitting objects, are addressed. Test results are presented and discussed. The algorithm may be used in the system...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Mispronunciation Detection in Non-Native (L2) English with Uncertainty Modeling

    Publikacja

    - Rok 2021

    A common approach to the automatic detection of mispronunciation in language learning is to recognize the phonemes produced by a student and compare it to the expected pronunciation of a native speaker. This approach makes two simplifying assumptions: a) phonemes can be recognized from speech with high accuracy, b) there is a single correct way for a sentence to be pronounced. These assumptions do not always hold, which can result...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Pedestrian Safety in Road Traffic in Poland

    Every third road accident in Poland involves a pedestrian as a participant or, most of the time, a casualty. Pedestrian accidents are usually the result of complex situations and the outcome of a number of factors related to driver and pedestrian behaviour and road infrastructure. Safety depends largely on how well the traffic condition is perceived and on visibility in traffic. The paper presents the results of analyses of methodologies...

    Pełny tekst do pobrania w portalu

  • Effects of Road Infrastructure on Pedestrian Safety

    The objective of the work was to identify risks for pedestrians that involve road infrastructure and roadside and to define how selected elements of geometry and traffic layout affect driver behaviour (speed on approaching pedestrian crossings). The results have helped to formulate recommendations on pedestrian crossing design. The research included an analysis of 2013-2017 statistics to identify the...

    Pełny tekst do pobrania w portalu

  • Accuracy of Differential Phase Delay Estimation for GPS Spoofing Detection

    Publikacja

    - Rok 2013

    GPS spoofing is an attack based on transmission of false signals to target receivers, in order to force the computation of incorrect time, position or velocity information. It is a threat which is recently considered with continuously growing awareness. There is a need for effective way of GPS spoofing detection. Many groups of methods are proposed in the literature. Spatial processing methods are considered to be robust in a wide...

  • Multiscaled Hybrid Features Generation for AdaBoost Object Detection

    This work presents the multiscaled version of modified census features in graphical objects detection with AdaBoost cascade training algorithm. Several experiments with face detector training process demonstrate better performance of such features over ordinal census and Haar-like approaches. The possibilities to join multiscaled census and Haar features in single hybrid cascade of strong classifiers are also elaborated and tested....

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Pedestrian safety management using the risk-based approach

    The paper presents a concept of a multi-level pedestrian safety management system. Three management levels are distinguished: strategic, tactical and operational. The basis for the proposed approach to pedestrian safety management is a risk-based method. In the approach the elements of behavioural and systemic theories were used, allowing for the development of a formalised and repeatable procedure integrating the phases of risk...

    Pełny tekst do pobrania w portalu

  • Empirical analyses of robustness of the square Msplit estimation

    Publikacja

    - Journal of Applied Geodesy - Rok 2021

    The paper presents Msplit estimation as an alternative to methods in the class of robust M-estimation. The analysis conducted showed that Msplit estimation is highly efficient in the identification of observations encumbered by gross errors, especially those of small or moderate values. The classical methods of robust estimation provide then unsatisfactory results. Msplit estimation also shows high robustness to single gross errors...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Sonar Pulse Detection Using Chirp Rate Estimation and CFAR Algorithms

    Publikacja

    This paper presents a new approach to sonar pulse detection. The method uses chirp rate estimators and algorithms for the adaptive threshold, commonly used in radiolocation. The proposed approach allows detection of pulses of unknown parameters, which may be used in passive hydrolocation or jamming detection in underwater communication. Such an analysis is possible thanks to a new kind of imaging, which presents signal energy in...

    Pełny tekst do pobrania w portalu

  • Resolving conflicts in object tracking for automatic detection of events in video

    Publikacja

    W referacie przedstawiono algorytm rozwiązywania konfliktów w śledzeniu obiektów ruchomych. Proponowana metoda wykorzystuje predykcję stanu obiektu obliczaną przez filtry Kalmana oraz dopasowuje wykryte obiekty do struktur śledzących ich ruch na podstawie deskryptorów koloru i tekstury. Omówiono specyficzne sytuacje powodujące konflikty, takie jak rozdzielanie obiektów. Przedstawiono wyniki testów. Algorytm może być zastosowany...

  • Detection and size estimation of crack in plate based on guided wave propagation

    The paper presents results of the comprehensive theoretical and experimental investigation of crack detection in metallic plate using guided wave propagation. The main aim of the paper is to develop the novel method which would allow for linear crack size estimation with the use of minimal number of the transducers. In general, there exists the relation between length of the propagation path and the wave amplitude value. However,...

    Pełny tekst do pobrania w portalu

  • Risk assessment methodologies for pedestrian crossings without traffic lights – Warsaw case study – pedestrian safety assessment

    Publikacja

    - MATEC Web of Conferences - Rok 2017

    Based on data from 2015 [1], it was found that 31% of all fatalities in road accidents in Poland were pedestrians. In places accessible to pedestrian traffic 74.2% of total accidents involving pedestrians were recorded. Approximately 53.9% of accidents involving pedestrians take place in the proximity of pedestrian crossings. In the context of improving the safety of vulnerable road users, an assessment of the condition of infrastructure...

    Pełny tekst do pobrania w portalu

  • The Influence of Camera and Optical System Parameters on the Uncertainty of Object Location Measurement in Vision Systems

    The article presents the influence of the camera and its optical system on the uncertainty of object position measurement in vision systems. The aim of the article is to present the methodology for estimating the combined standard uncertainty of measuring the object position with a vision camera treated as a measuring device. The identification of factors affecting the location measurement uncertainty and the determination of their...

    Pełny tekst do pobrania w portalu

  • Pedestrian safety in road traffic – studies, recommendations and proposed improvements

    Pedestrians are involved and most frequently the victims of every third road accident in Poland. Pedestrian accidents most frequently occur in complex circumstances, as a result of many factors related to the behaviour of drivers and pedestrians. The basic parameters that determine road safety include the perception of traffic and visibility on the road. The paper will present the results of research conducted within the project...

    Pełny tekst do pobrania w portalu

  • Detection and Direction-of-Arrival Estimation of Weak Spread Spectrum Signals Received with Antenna Array

    Publikacja

    - Electronics - Rok 2021

    This paper presents a method for the joint detection and direction of arrival (DOA) estimation of low probability of detection (LPD) signals. The proposed approach is based on using the antenna array to receive spread-spectrum signals hidden below the noise floor. Array processing exploits the spatial correlation between phase-delayed copies of the signal and allows us to evaluate the parameter used to make the decision about the...

    Pełny tekst do pobrania w portalu

  • CANCER DETECTION AND PREVENTION

    Czasopisma

    ISSN: 0361-090X

  • Biomolecular Detection and Quantification

    Czasopisma

    eISSN: 2214-7535

  • Fully Automated AI-powered Contactless Cough Detection based on Pixel Value Dynamics Occurring within Facial Regions

    Publikacja

    - Rok 2021

    Increased interest in non-contact evaluation of the health state has led to higher expectations for delivering automated and reliable solutions that can be conveniently used during daily activities. Although some solutions for cough detection exist, they suffer from a series of limitations. Some of them rely on gesture or body pose recognition, which might not be possible in cases of occlusions, closer camera distances or impediments...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Performance Evaluation of Selected Parallel Object Detection and Tracking Algorithms on an Embedded GPU Platform

    Publikacja

    - Rok 2017

    Performance evaluation of selected complex video processing algorithms, implemented on a parallel, embedded GPU platform Tegra X1, is presented. Three algorithms were chosen for evaluation: a GMM-based object detection algorithm, a particle filter tracking algorithm and an optical flow based algorithm devoted to people counting in a crowd flow. The choice of these algorithms was based on their computational complexity and parallel...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Determining and verifying the safety integrity level of the safety instrumented systems with the uncertainty and security aspects

    Safety and security aspects consist of two different group of functional requirements for the control and protection systems. In the paper it is proposed that the security analysis results can be used as a factor increasing or decreasing the risk level. It concerns a process of determining required safety integrity level of given safety functions. The authors propose a new approach for functional safety risk analysis. In this case...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Closer Look at the Uncertainty Estimation in Semantic Segmentation under Distributional Shift

    While recent computer vision algorithms achieve impressive performance on many benchmarks, they lack robustness - presented with an image from a different distribution, (e.g. weather or lighting conditions not considered during training), they may produce an erroneous prediction. Therefore, it is desired that such a model will be able to reliably predict its confidence measure. In this work, uncertainty estimation for the task...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • A survey of neural networks usage for intrusion detection systems

    In recent years, advancements in the field of the artificial intelligence (AI) gained a huge momentum due to the worldwide appliance of this technology by the industry. One of the crucial areas of AI are neural networks (NN), which enable commer‐ cial utilization of functionalities previously not accessible by usage of computers. Intrusion detection system (IDS) presents one of the domains in which neural networks are widely tested...

    Pełny tekst do pobrania w portalu

  • Objects classification based on their physical sizes for detection of events in camera images

    Publikacja

    In the paper, a method of estimation of the physical sizes of the objects tracked in the video surveillance system, and a simple module for object classification based on the estimated physical sizes, are presented. The results of object classification are then used for automatic detection of various types of events in the camera image.

  • Radiation Detection Technology and Methods

    Czasopisma

    ISSN: 2509-9930 , eISSN: 2509-9949

  • Uncertainty estimation of loop impedance measurement determined by the vector method

    This article presents a detailed analysis of uncertainty estimation of loop impedance measurement determined by the vector method. The analysis includes the following estimates: resistance variance, voltage variance and time measurement variance. This paper presents a methodology for estimating the combined standard uncertainty of loop impedance by the vector method. The vector method allows to determine loop impedance based on...

  • Comparison of edge detection algorithms for electric wire recognition

    Publikacja

    Edge detection is the preliminary step in image processing for object detection and recognition procedure. It allows to remove useless information and reduce amount of data before further analysis. The paper contains the comparison of edge detection algorithms optimized for detection of horizontal edges. For comparison purposes the algorithms were implemented in the developed application dedicated to detection of electric line...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Direct detection of quantum entanglement

    Publikacja

    - PHYSICAL REVIEW LETTERS - Rok 2002

    Basing on positive maps separability criterion we propose the experimentally viable, direct detection of quantum entanglement. It is efficient and does not require any a priori knowledge about the state. For two qubits it provides a sharp (i.e., “if and only if”) separability test and estimation of amount of entanglement. We view this method as a new form of quantum computation, namely, as a decision problem with quantum data structure.

  • Instrument detection and pose estimation with rigid part mixtures model in video-assisted surgeries

    Localizing instrument parts in video-assisted surgeries is an attractive and open computer vision problem. A working algorithm would immediately find applications in computer-aided interventions in the operating theater. Knowing the location of tool parts could help virtually augment visual faculty of surgeons, assess skills of novice surgeons, and increase autonomy of surgical robots. A surgical tool varies in appearance due to...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Automatic detection of abandoned luggage employing a dual camera system

    Publikacja

    - Rok 2010

    A system for automatic detection of events using a system of fixed and PTZ (pan-tilt-zoom) cameras is described. Images from the fixed camera are analyzed by means of object detection and tracking. Event detection system uses a set of rules to analyze data on the tracked moving objects and to detect defined events. A PTZ camera is used to obtain a detailed view of a selected object. A procedure for conversion between the pixel...

  • A framework for automatic detection of abandoned luggage in airport terminal

    Publikacja

    A framework for automatic detection of events in a video stream transmitted from a monitoring system is presented. The framework is based on the widely used background subtraction and object tracking algorithms. The authors elaborated an algorithm for detection of left and removed objects based on mor-phological processing and edge detection. The event detection algorithm collects and analyzes data of all the moving objects in...

  • Sensorless Fault Detection of Induction Motor with Inverter Output Filter

    Publikacja

    The paper presents the problem of monitoring and fault detection of a sensorless voltage inverter fed squirrel cage induction motor with LC filter. The detection is based on load torque estimation of the investigated torque transmission system. The load torque is calculated besides the computation of other variables that are mandatory for sensorless drive operation such as rotor flux and speed. The implemented LC filter smooths...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Detection of damages in a rivetted plate

    Publikacja

    The paper presents the results of damage detection in a riveted aluminium plate. The detection method has been based on Lamb wave propagation. The plate has been analysed numerically and experimentally. Numerical calculations have been carried out by the use of the time-domain spectral finite element method, while for the experimental analysis laser scanning Doppler vibrometry (LSDV) has been utilised. The panel has been excited...

  • Determining and verifying the safety integrity level of the controland protection systems under uncertainty

    Publikacja

    The probabilistic models of the electrical, electronic and programmable electronic (E/E/PE)systems, which are proposed in the IEC 61508 and IEC 61511 standards, do not cover the uncertainty aspectsof results obtained from these models. The proposed method takes into consideration the sensitivity analysisof probabilistic models of E/E/PE systems as well as the uncertainty of probabilistic results. It uses differentialfactors, which...

  • A novel concept for tissue-metal detection and differentiation using an inductive proximity sensor

    Publikacja

    - Rok 2016

    In this paper a novel application of inductive proximity sensors for detection of living tissue by means of measurements of the coil impedance changes at different frequencies is described. The mathematical analyses utilizing Bessel function estimation include detected object size and its distance from a sensor. The main aim of this study is to prove the possibility of distinguishing between metal objects and living tissues. The...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Modular machine learning system for training object detection algorithms on a supercomputer

    Publikacja

    - Rok 2010

    W pracy zaprezentowano architekturę systemu służącego do tworzenia algorytmów wykorzystujących metodę AdaBoost i służących do wykrywania obiektów (np. twarzy) na obrazach. System został podzielony na wyspecjalizowane moduły w celu umożliwienia łatwej rozbudowy i efektywnego zrównoleglenia implementacji przeznaczonej dla superkomputera. Na przykład, system może być rozszerzony o nowe cechy i algorytmy ich ekstrakcji bez konieczności...

  • Detection of vehicles stopping in restricted zones in video from surveillance cameras

    Publikacja

    - Rok 2014

    An algorithm for detection of vehicles that stop in restricted areas, e.g. excluded by traffic rules, is proposed. Classic approaches based on object tracking are inefficient in high traffic scenes because of tracking errors caused by frequent object merging and splitting. The proposed algorithm uses the background subtraction results for detection of moving objects, then pixels belonging to moving objects are tested for stability....

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Transient detection for speech coding applications

    Signal quality in speech codecs may be improved by selecting transients from speech signal and encoding them using a suitable method. This paper presents an algorithm for transient detection in speech signal. This algorithm operates in several frequency bands. Transient detection functions are calculated from energy measured in short frames of the signal. The final selection of transient frames is based on results of detection...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Faults and Fault Detection Methods in Electric Drives

    Publikacja

    The chapter presents a review of faults and fault detection methods in electric drives. Typical faults are presented that arises for the induction motor, which is valued in the industry for its robust construction and cost-effective production. Moreover, a summary is presented of detectable faults in conjunction with the required physical information that allow a detection of specific faults. In order to address faults of a complete...

    Pełny tekst do pobrania w serwisie zewnętrznym

  • Detection and segmentation of moving vehicles and trains using Gaussian mixtures, shadow detection and morphological processing

    Publikacja

    Solution presented in this paper combines background modelling, shadow detection and morphological and temporal processing into one system responsible for detection and segmentation of moving objects recorded with a static camera. Vehicles and trains are detected based on their pixellevel difference from the continually updated background model utilizing a Gaussian mixture calculated separately for every pixel. The shadow detection...

  • Layered background modeling for automatic detection of unattended objects in camera images

    Publikacja

    - Rok 2011

    An algorithm for automatic detection of unattended objects in video camera images is presented. First, background subtraction is performed, using an approach based on the codebook method. Results of the detection are then processed by assigning the background pixels to time slots, based on the codeword age. Using this data, moving objects detected during a chosen period may be extracted from the background model. The proposed approach...

    Pełny tekst do pobrania w serwisie zewnętrznym