ARM-Based Video Intercom System with Next-Gen Human Presence Detection using Deep Learning : magistrsko deloMario Gavran
, 2022, magistrsko delo
Opis: This master's thesis presents an advanced video system with human presence detection based on deep learning and an ARM microcontroller. The objective of the thesis is to develop a system that works as a smart video intercom, which could be installed, e.g. on the entrance door, and autonomously alert the owner that a guest is in front of the door. The main goal is to use an AI algorithm, namely the neural network model on a constrained device, such as an ARM microcontroller, as their main advantage is lower power consumption and cost.
The thesis also describes commonly used methods to reduce the power and memory footprint and to implement and accelerate the deep learning algorithms more effectively. Further, the most notable deep learning hardware and some general platforms are described in more detail.
The thesis also presents the development of a human presence detection system based on an ARM microcontroller, VGA camera, and LCD, where Tensorflow Lite Micro, an open-source C++ framework for deploying deep learning models to embedded platforms and a pre-trained neural network model for person presence detection are used.
Ključne besede: TensorFlow Lite Micro, Video intercom system, ARM Cortex-M microcontroller, Human presence detection, Neural network
Objavljeno v DKUM: 08.07.2022; Ogledov: 63; Prenosov: 19
Celotno besedilo (7,86 MB)
Development of an automated measurement system for UWB Radar : master's thesisTomaž Leopold
, 2022, magistrsko delo
Opis: Ultra-wideband is becoming a key technological element in the consumer industry, providing a safe and reliable way to enable more precise location services as well as secure access to a multitude of systems. Ultra-wideband radar is a subset of that technology, providing not only device-free localization techniques but also complementing expensive medical-level devices or providing means to measure vital signs in a more accessible way. Ultra-wideband radars embedded in smart devices like smartphones can be used for presence detection, gesture recognition, and vital sign monitoring (i.e., breathing detection and heart rate monitoring). Ultra-wideband radar is still in development, and during that process it needs to be well-tested to match the customers’ expectations and standards that apply to the technology. In this master’s thesis, we start by providing a brief overview of ultra-wideband technology. Then we propose and develop an automated measurement system for UWB radar that tests the radio frequency functionality of a radar device, its compliance with standards, frequency regulations, and use-case testing. Performing measurements to test the aforementioned radar devices is a time-consuming and cumbersome task that can be replaced by our developed automated measurement system. We demonstrate the advantages in terms of reduced measurement time and improved test reproducibility, resulting in a well-tested ultra-wideband radar device during the development cycle.
Ključne besede: automated system, channel impulse response, IR-UWB, radar
Objavljeno v DKUM: 17.02.2022; Ogledov: 385; Prenosov: 0
Systemic direction and supervision of police work in the Republic of Slovenia : current situation analysisMaja Modic
, 2020, izvirni znanstveni članek
Opis: Purpose: The purpose of this paper is to present the findings of the target research project "Effectiveness of Systemic Supervision of the Police in Regard to Human Rights as well as the Statutory and Professional Standards of Police Work (V5-1942)". We focused on the implementation of systemic supervision of police work, more precisely, on the operations of the Division for System Guidelines and Supervision of the Police, which operates within the Police and Security Directorate. Design/Methods/Approach: We examined how systemic supervisory activity of the work of the police is regulated by Slovenian law, limiting ourselves to the work of the Division for System Guidelines and Supervision of the Police. Based on the findings and an additional review of anonymised supervision reports, we prepared a questionnaire and conducted a structured written interview as well as oral interviews with the Division's employees. Findings: Systemic supervision efforts are contributing significantly to the quality of policing, but challenges and opportunities for improvement still remain. One of the main challenges is keeping police officers informed about supervision reports and effectively implementing inspection findings. Research Limitations/Implication: In the part of the study described, we focus exclusively on the work of the Division for System Guidelines and Supervision of the Police and take a look at systemic supervision activities solely from the Division's point of view. For a more complete overview of the matter, our findings should be compared with the perspective of those being supervised, i.e. the police officers who have undergone supervision. Originality/Value: The main added value of the paper is that it examines how supervisors perceive their own work and the organisation, operation and effectiveness of systemic supervisions of police work.
Ključne besede: systemic supervision of police work, Division for System Guidelines and Supervision of the Police, Police and Security Directorate, Slovenia
Objavljeno v DKUM: 11.06.2021; Ogledov: 424; Prenosov: 14
Povezava na celotno besedilo
Strategy of the University of Maribor : 2021-2030
Opis: The strategy of the University of Maribor pursues an inclusive, innovative and integrated higher education area that will train active, critically-thinking and responsible citizens, guarantee the quality of education and research, academic integrity, and cultivate concern for sustainable development of society. It stresses the importance of the freedom of research and institutional autonomy, development of lifelong learning, digitalisation and green infrastructure. It follows the orientations of the Rome Ministerial Communiqué of 2020 on the development of higher education in Europe, the Magna Charta Universitatum 2020 principles and the national strategic documents on higher education.
Ključne besede: organization and connectivity, education, scientific research and art, stimulating working environment, students, internationalisation, development through quality system, involvement in local environment, spatial development, IT support
Objavljeno v DKUM: 14.04.2021; Ogledov: 347; Prenosov: 25
Povezava na datoteko
The need for a unified patent courtEugénio Lucas
, 2020, izvirni znanstveni članek
Opis: This article discusses the need of a Unified Patent Court, as a major contribution for the implementation of the European unitary patent. It begins by outlining the evolution of European patent rights and identifying the problems of the current system. Implementing a Unified Patent Court will be defended as a solution for one of the most important problems of the European patent: the lack of a central mechanism of jurisdictional control. The current model of jurisdictional control gives rise to a situation of great legal uncertainty and has diminished the value of European Union patent rights. We will analyse the virtues and disadvantages of the new European Unitary Patent resulting from the adoption of Regulations (EU) 1257/2012 and 1260/2012 and the Agreement on a Unified Patent Court.
Ključne besede: unified patent court, European unitary patent, European patent convention, patent itigation, European patent system
Objavljeno v DKUM: 15.01.2021; Ogledov: 352; Prenosov: 0
The standards of proof in medical malpractice casesNina Cek
, 2020, izvirni znanstveni članek
Opis: The article examines the procedural aspect of medical malpractice cases. It focuses on the differences in standards of proof by first explaining the characteristics of the Slovenian legal system and then comparing it with German and English legal systems. The author sheds light on the approach of the EU court on the question of the responsibility of the manufacturer for the product (vaccine) and suggests the direction to use a broader framework for the evaluation of evidence and presumptions. Given the disclosed problems of proving through the help of a medical expert, the article emphasizes the importance of respect for human rights in civil proceedings. Particular emphasis is also placed on no-fault systems and the question is raised of how the introduction of such a system into the Slovenian legal system would affect the perceived problem of proving a medical error and informed consent omission.
Ključne besede: medical malpractice, expert evidence, standards of proof, no-fault system, civil procedure rules
Objavljeno v DKUM: 15.01.2021; Ogledov: 281; Prenosov: 23
Povezava na datoteko
Energy efficient system for detection of elephants with Machine Learning : master's thesisMarko Sagadin
, 2020, magistrsko delo
Opis: Human-Elephant Conflicts are a major problem in terms of elephant conservation.
According to WILDLABS, an average of 400 people and 100 elephants are killed every year in India alone because of them.
Early warning systems replace the role of human watchers and warn local communities of nearby, potentially life threatening, elephants, thus minimising the Human-Elephant Conflicts.
In this Master's thesis we present the structure of an early warning system, which consists of several low-power embedded systems equipped with thermal cameras and a single gateway.
To detect elephants from captured thermal images we used Machine Learning methods, specifically Convolutional Neural Networks.
The main focus of this thesis was the design, implementation and evaluation of Machine Learning models running on microcontrollers under low-power conditions.
We designed and trained several accurate image classification models, optimised them for on-device deployment and compared them against models trained with commercial software in terms of accuracy, inference speed and size.
While writing firmware, we ported a part of the TensorFlow library and created our own build system, suitable for the libopencm3 platform.
We also implemented reporting of inference results over the LoRaWAN network and described a possible server-size solution.
We finally a constructed fully functional embedded system from various development and evaluation boards, and evaluated its performance in terms of power consumption.
We show that embedded systems with Machine Learning capabilities are a viable solution to many real life problems.
Ključne besede: machine learning, microcontroller, on-device inference, thermal camera, low-power system
Objavljeno v DKUM: 06.01.2021; Ogledov: 523; Prenosov: 99
Celotno besedilo (13,35 MB)
Deep Learning on Low Power Embedded Devices Using RISC-V Cores with an Extended Instruction Set : master's thesisJure Vreča
, 2020, magistrsko delo
Opis: This thesis explores the possibility of running neural networks on microcontrollers and how
to optimize their performance using instruction set extensions. Microcontrollers are seen as
too weak to run neural networks. We challenge this view and show that stripped-down neural
networks can run and be useful for some applications. We used an open-source microcontroller
called PULPino to run our neural network. The benefit of various instructions and optimizations
for minimizing energy consumption to run deep learning algorithms was evaluated. Hardware
loops, loop unrolling, and the dot-product unit were implemented and tested.
We developed an FPGA-based testing system to evaluate our hardware. We also developed a
deep learning library and a test neural network for our hardware. We wrote two versions of
the deep learning library. One version is the reference code, and the other is the optimized
code that uses the dot product unit. Using the testing system, we tested the performance of
the two versions. The synthesis was run to determine the power and energy consumption. We
also tried out various optimizations to see if the performance could be improved.
Using instruction set extensions and algorithmic optimizations we reduced the clock cycle count by 72% for the convolutional layers and by 78% for fully-connected layers. This reduced
power consumption by 73%. We compare our results with related research.
Ključne besede: deep learning, embedded system, instruction set, RISC-V
Objavljeno v DKUM: 03.11.2020; Ogledov: 773; Prenosov: 127
Celotno besedilo (2,69 MB)
Optimization of the distribution network operation by integration of distributed energy resources and participation of active elements : doctoral thesisNevena Srećković
, 2020, doktorska disertacija
Opis: Distribution Networks (DNs) are evolving from a once passive to an active part of the electricity network. This evolution is driven by the current political and environmental decisions, Directives and Incentives, as well as the technological development, observed in the everincreasing integration of renewable energy resources, advanced network control and measurement devices, the upcoming energy exibility market, etc. This Doctoral Thesis deals with the problem of optimization of the technical aspects of a DN operation, enabled by the proliferated integration of the photovoltaic systems (PV) and other active devices. The main objective of the Thesis is the optimization of a DN operation in terms of minimization of electrical energy losses while ensuring the proper voltage profiles and preventing thermal overloadingof lines. Therefore, three Differential Evolution-based optimization procedures were developed and tested on real medium and low voltage DNs. The first methodology determines the optimal rooftop surfaces for the installation of PV systems, yielding minimum annual energy losses. It is based on the simultaneous consideration of high-resolution spatio-temporal solar and PV potential data, as well as long-term measured profiles of consumption and generation of electrical energy within the network of a given configuration. The second algorithm minimizes network losses in a time-discrete operation point, by determining the optimal operation of the following active elements: PV systems capable of cooperation in reactive power provision, On-Load Tap Changer equipped transformer substations and remotely controlled switches for network reconfiguration. The final algorithm was developed by a proper consolidation of the first two approaches, yielding the synergistic effects expressed as the increase of loss reduction and network exibility. The results of the performed case studies show that the locations of the highest suitability for PV installation with respect to the solar energy availability, are not necessarily the best choice from the network operation standpoint. Therefore, both standpoints should be considered simultaneously when choosing the rooftop surfaces for PV installation. Furthermore, by determining optimal hourly operation of the considered active elements, not only the additional reduction of annual network losses was achieved, but also increased accommodation of the PV systems that doesn't violate operation constraints.
Ključne besede: distribution network, optimization of operation, active network elements, PV system placement, minimization of losses
Objavljeno v DKUM: 11.06.2020; Ogledov: 1077; Prenosov: 241
Celotno besedilo (44,58 MB)