Wyniki wyszukiwania - BazTech

1

Event Detection System for the Penitentiary Institutions using Multimodal Data and Deep Networks

Bilski Piotr, Lewandowski Marcin, Bilski Adrian, Buchowicz Andrzej, Olejnik Jacek, Mazurek Paweł, Jędrzejewski Konrad

Machine Graphics and Vision

|

2023

|

Vol. 32, No. 3/4

233--254

EN

The aim of the paper is to present the distributed system for the unwanted event detection regarding inmates in the closed penitentiary facilities. The system processes large number of data streams from IP cameras (up to 180) and performs the event detection using Deep Learning neural networks. Both audio and video streams are processed to produce the classification outcome. The application-specific data set has been prepared for training the neural models. For the particular event types 3DCNN and YOLO architectures have been used. The system was thoroughly tested both in the laboratory conditions and in the actual facility. Accuracy of the particular event detection is on the satisfactory level, though problems with the particular events have been reported and will be dealt with in the future.

2

Modelowe rozwiązania skanerów UTPA do badań spawów dla wież wiatrowych, sekcji płaskich oraz konstrukcji wielkogabarytowych on-shore/off-shore

Lewandowski Marcin, Rozbicki Jakub, Smach Hanna, Karwat Piotr, Szczurek Arkadiusz, Sala Jolanta, Bera Alicja

Badania Nieniszczące i Diagnostyka

|

2023

|

nr 1-4

97--100

PL

W ramach realizowanego projektu wdrożeniowego (akronim: BalTECH, finansowanie NCBR POIR) opracowano modelowe stanowiska skanerów UTPA do badań nieniszczących spawów dla asortymentu produktów wytwarzanych w Baltic Operator sp. z o.o. Skanery zapewniają prowadzenie i sprzężenie dwóch głowic Phased-Array (badanie dwustronne). Do realizacji badań UTPA wykorzystano komercyjny aparat Olympus-OmniScan™ X3, natomiast dla metody UTPA-FMC (Full-Matrix Capture) badawczą platformę ultradźwiękową us4R-lite™ firmy us4us sp. z o.o. Wykonano zestaw ok. 170 próbek testowych spawów z różnymi niezgodnościami dla płyt w zakresie grubości 12–65 mm, które zostały przebadanie metodami VT, MT/PT, UT, RT, UTPA. Opracowana procedura badania i wzorce testowe pozwoliły na pełną walidację klasycznej metody UTPA do badania sekcji wież wiatrowych. Eksperymentalne zastosowanie i porównanie metody UTPA-FMC pokazało jej duży potencjał oraz nowe możliwości wizualizacji i oceny wad, w stosunku do klasycznej metody UTPA. Zweryfikowano także możliwość zbierania surowych danych FMC z prędkością do 100 mm/s. Kluczowe znaczenie ma wdrożenie nowoczesnych i ekonomicznych rozwiązań badań nieniszczących, które zapewnią ocenę jakości 100% długości spawu. Istotny wkład w rozwój laboratoriów badawczych, w kontekście wiarygodności uzyskiwanych wyników badania.

EN

As part of an ongoing project (acronym: BalTECH, NCBR POIR funding), model UTPA scanner stations were developed for nondestructive testing of welds for a range of products manufactured at Baltic Operator Ltd. The scanners provide guidance and coupling of two Phased-Array probes (two-sided testing). A commercial Olympus-OmniScan™ X3 apparatus was used for the UTPA testing, while for the UTPA-FMC method the us4R-lite™ ultrasound research platform from us4us sp. z o.o. was used. A set of about 170 weld test specimens with various nonconformities for plates in the thickness range of 12–65 mm was prepared and tested by VT, MT/PT, UT, RT, UTPA methods. The developed test procedure and test patterns allowed full validation of the classical UTPA method for testing wind tower sections. The experimental application and comparison of the UTPA-FMC method showed its great potential and new possibilities for visualization and evaluation of defects, compared to the classical UTPA method. The ability to collect raw FMC data at speeds of up to 100 mm/s was also verified. The goal of the project is to implement modern and cost effective nondestructive testing solutions that will provide quality assessment of 100% of the weld length.

3

EMD-based time-frequency analysis methods of non-stationary audio signals

Lewandowski Marcin, Grodzicka Salomea

Vibrations in Physical Systems

|

2022

|

Vol. 33, nr 2

art. no. 2022215

EN

To ensure that any time series data is appropriately interpreted, it should be analyzed with proper signal processing tools. The most common analysis methods are kernel-based transforms, which use base functions and their modifications to represent time series data. This work discusses an analysis of audio data and two of those transforms - the Fourier transform and the wavelet transform based on a priori assumptions about the signal's linearity and stationarity. In audio engineering, these assumptions are invalid because the statistical parameters of most audio signals change with time and cannot be treated as an output of the LTI system. That is why recent approaches involve decomposition of a signal into different modes in a data-dependent and adaptive way, which may provide advantages over kernel-based transforms. Examples of such methods include empirical mode decomposition (EMD), ensemble EMD (EEMD), variational mode decomposition (VMD), or singular spectrum analysis (SSA). Simulations were performed with speech signal for kernel-based and data-dependent decomposition methods, which revealed that evaluated decomposition methods are promising approaches to analyzing non-stationary audio data.

4

Modelowe rozwiązania skanerów UTPA do badań spawów dla wież wiatrowych sekcji płaskich oraz konstrukcji wielkogabarytowych on-shore/off-shore

Lewandowski Marcin, Rozbicki Jakub, Smach Hanna, Karwat Piotr, Szczurek Arkadiusz, Sala Jolanta, Bera Alicja

Badania Nieniszczące i Diagnostyka

|

2022

|

1-4

89--92

PL

W ramach realizowanego projektu wdrożeniowego (akronim: BalTECH, finansowanie NCBR POIR) opracowano modelowe stanowiska skanerów UTPA do badań nieniszczących spawów dla asortymentu produktów wytwarzanych w Baltic Operator sp. z o.o. Skanery zapewniają prowadzenie i sprzężenie dwóch głowic Phased-Array (badanie dwustronne). Do realizacji badań UTPA wykorzystano komercyjny aparat Olympus-OmniScan™ X3, natomiast dla metody UTPA-FMC (Full-Matrix Capture) badawczą platformę ultradźwiękową us4R-lite™ firmy us4us sp. z o.o. Wykonano zestaw ok. 170 próbek testowych spawów z różnymi niezgodnościami dla płyt w zakresie grubości 12–65 mm, które zostały przebadanie metodami VT, MT/PT, UT, RT, UTPA. Opracowana procedura badania i wzorce testowe pozwoliły na pełną walidację klasycznej metody UTPA do badania sekcji wież wiatrowych. Eksperymentalne zastosowanie i porównanie metody UTPA-FMC pokazało jej duży potencjał oraz nowe możliwości wizualizacji i oceny wad, w stosunku do klasycznej metody UTPA. Zweryfikowano także możliwość zbierania surowych danych FMC z prędkością do 100 mm/s. Kluczowe znaczenie ma wdrożenie nowoczesnych i ekonomicznych rozwiązań badań nieniszczących, które zapewnią ocenę jakości 100% długości spawu. Istotny wkład w rozwój laboratoriów badawczych, w kontekście wiarygodności uzyskiwanych wyników badania.

EN

As part of an ongoing project (acronym: BalTECH, NCBR POIR funding), model UTPA scanner stations were developed for nondestructive testing of welds for a range of products manufactured at Baltic Operator Ltd. The scanners provide guidance and coupling of two Phased-Array probes (two-sided testing). A commercial Olympus-Omni Scan™ X3 apparatus was used for the UTPA testing, while for the UTPA-FMC method the us4R-lite™ ultrasound research platform from us4us sp. z o.o. was used. A set of about 170 weld test specimens with various nonconformities for plates in the thickness range of 12–65 mm was prepared and tested by VT, MT/PT, UT, RT, UTPA methods. The developed test procedure and test patterns allowed full validation of the classical UTPA method for testing wind tower sections. The experimental application and comparison of the UTPA-FMC method showed its great potential and new possibilities for visualization and evaluation of defects, compared to the classical UTPA method. The ability to collect raw FMC data at speeds of up to 100 mm/s was also verified. The goal of the project is to implement modern and cost effective nondestructive testing solutions that will provide quality assessment of 100% of the weld length.

5

Breast lesion classification based on ultrasonic radio-frequency signals using convolutional neural networks

Jarosik Piotr, Klimonda Ziemowit, Lewandowski Marcin, Byra Michal

Biocybernetics and Biomedical Engineering

|

2020

|

Vol. 40, no. 3

977--986

EN

We propose a novel approach to breast mass classification based on deep learning models that utilize raw radio-frequency (RF) ultrasound (US) signals. US images, typically displayed by US scanners and used to develop computer-aided diagnosis systems, are reconstructed using raw RF data. However, information related to physical properties of tissues present in RF signals is partially lost due to the irreversible compression necessary to make raw data readable to the human eye. To utilize the information present in raw US data, we develop deep learning models that can automatically process small 2D patches of RF signals and their amplitude samples. We compare our approach with classification method based on the Nakagami parameter, a widely used quantitative US technique utilizing RF data amplitude samples. Our better performing deep learning model, trained using RF signals and their envelope samples, achieved good classification performance, with the area under the receiver attaining operating characteristic curve (AUC) and balanced accuracy of 0.772 and 0.710, respectively. The proposed method significantly outperformed the Nakagami parameter-based classifier, which achieved AUC and accuracy of 0.64 and 0.611, respectively. The developed deep learning models were used to generate parametric maps illustrating the level of mass malignancy. Our study presents the feasibility of using RF data for the development of deep learning breast mass classification models.

6

Automatic recognition of artificial reverberation settings in speech recordings

Kachniarz Krzysztof, Lewandowski Marcin

Vibrations in Physical Systems

|

2019

|

Vol. 30, nr 1

art. no. 2019125

EN

The aim of this study is to create the method for automatic recognition of artificial reverberation settings extracted from a reference speech recordings. The proposed method employs machine-learning techniques to support the sound engineer in finding the ideal settings for artificial reverberation plugin available at a given Digital Audio Workstation (DAW), i.e. Gaussian Mixture Model (GMM) approach and deep Convolutional Neural Network (CNN) VGG13, which is a novel approach. Training set and data set are 1885 speech signals selected from a EMIME Bilingual Database which were processed with 66 artificial reverberation presets selected from Semantic Audio Labs’s SAFE Reverb plugin database. Performance of the proposed automatic recognition method was evaluated using similarity measures between features of reference and analysed speech recordings. Evaluation procedure showed that a classical GMM approach gives 43.8% of recognition accuracy while proposed method with VGG13 deep CNN gives 99.94% of accuracy.

7

Optimization of Short-Lag Spatial Coherence Imaging Method

Domaradzki Jakub, Lewandowski Marcin, Żołek Norbert, Lewandowski Marcin

Archives of Acoustics

|

2019

|

Vol. 44, No. 4

669--679

EN

The computing performance optimization of the Short-Lag Spatial Coherence (SLSC) method applied to ultrasound data processing is presented. The method is based on the theory that signals from adjacent receivers are correlated, drawing on a simplified conclusion of the van Cittert-Zernike theorem. It has been proven that it can be successfully used in ultrasound data reconstruction with despeckling. Former works have shown that the SLSC method in its original form has two main drawbacks: time-consuming processing and low contrast in the area near the transceivers. In this study, we introduce a method that allows to overcome both of these drawbacks. The presented approach removes the dependency on distance (the “lag” parameter value) between signals used to calculate correlations. The approach has been tested by comparing results obtained with the original SLSC algorithm on data acquired from tissue phantoms. The modified method proposed here leads to constant complexity, thus execution time is independent of the lag parameter value, instead of the linear complexity. The presented approach increases computation speed over 10 times in comparison to the base SLSC algorithm for a typical lag parameter value. The approach also improves the output image quality in shallow areas and does not decrease quality in deeper areas.