| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

Search the digital library catalog Help

Query: search in
search in
search in
search in
* old and bologna study programme

Options:
  Reset


1 - 10 / 12
First pagePrevious page12Next pageLast page
1.
2.
A VAN-Based Multi-Scale Cross-Attention Mechanism for Skin Lesion Segmentation Network
Shuang Liu, Zeng Zhuang, Yanfeng Zheng, Simon Kolmanič, 2023, original scientific article

Abstract: With the rise of deep learning technology, the field of medical image segmentation has undergone rapid development. In recent years, convolutional neural networks (CNNs) have brought many achievements and become the consensus in medical image segmentation tasks. Although many neural networks based on U-shaped structures and methods, such as skip connections have achieved excellent results in medical image segmentation tasks, the properties of convolutional operations limit their ability to effectively learn local and global features. To address this problem, the Transformer from the field of natural language processing (NLP) was introduced to the image segmentation field. Various Transformer-based networks have shown significant performance advantages over mainstream neural networks in different visual tasks, demonstrating the huge potential of Transformers in the field of image segmentation. However, Transformers were originally designed for NLP and ignore the multidimensional nature of images. In the process of operation, they may destroy the 2D structure of the image and cannot effectively capture low-level features. Therefore, we propose a new multi-scale cross-attention method called M-VAN Unet, which is designed based on the Visual Attention Network (VAN) and can effectively learn local and global features. We propose two attention mechanisms, namely MSC-Attention and LKA-Cross-Attention, for capturing low-level features and promoting global information interaction. MSC-Attention is designed for multi-scale channel attention, while LKA-Cross-Attention is a cross-attention mechanism based on the large kernel attention (LKA). Extensive experiments show that our method outperforms current mainstream methods in evaluation metrics such as Dice coefficient and Hausdorff 95 coefficient.
Keywords: CNNs, deep learning, medical image processing, NLP, semantic segmentation
Published in DKUM: 14.03.2024; Views: 496; Downloads: 311
.pdf Full text (1,46 MB)
This document has many files! More...

3.
Comparison of different stator topologies for BLDC drives : master's thesis
Mitja Garmut, 2020, master's thesis

Abstract: The focus of this Master's thesis was to increase the output-power density of a fractional-horsepower BLDC drive. Different stator segmentation topologies were analyzed and evaluated for this purpose. The presented analysis was performed by using various models with different complexity levels, where a Magnetic Equivalent Circuit (MEC) model and a 2D transient Finite Element Method (FEM) model combined with a power-loss model, were applied systematically. Characteristic behavior of the BLDC drive was obtained in this way. The models were validated with measurement results obtained on an experimental test drive system. The influence of the weakening of the magnetic flux density and flux linkage, due to segmentation were analyzed based on the validated models. Furthermore, the increase of the thermal-stable output power and efficiency was rated, due to the consequently higher slot fill factor. Lastly, a detailed iron-loss analysis was performed for different stator topologies. The performed analysis showed that segmentation of the stator can enable a significant increase of the output power of the discussed BLDC drives, where the positive effects of segmentation outweigh the negative ones from the electromagnetic point of view. Segmentation, however, also impacts other domains, such as Mechanical and Thermal, which was out of the scope of this thesis, and will be performed in the future.
Keywords: fractional-horsepower BLDC drive, stator segmentation, fill factor increase, thermal-stable output power, Finite Element Method model
Published in DKUM: 17.11.2020; Views: 1373; Downloads: 38
.pdf Full text (1,69 MB)

4.
Work-Home conflict and strain: the role of work-related smartphone use, job insecurity and segmentation preferences
Tjaša Srnko, 2018, master's thesis

Abstract: Nowadays, through organisational and technological changes, organizations expect availability and work from their employees not just during working hours, but also in their free time. Work can be done from anywhere and at any time, also with the help of smartphones, which is one of the main constructs of the current research. If the employee is working from home on their smartphone, this can have an important contribution to managing work and private life, and also to strain. The individual’s preference of whether to segment his or her work from home also plays an important role in experiencing internal conflicts. As the motivation for work-related smartphone use is not fully clear, job insecurity was additionally explored as a potential predictor. In an online study, conducted in Austria, we collected data from 454 participants of different ages, gender and working backgrounds. For the analysis, multiple moderated regression and mediation analyses were used. Results showed significant relations between work-home interference and strain but no significant results between those constructs while adding work-related smartphone use. Segmentation preferences did not have an influence on the relationship between work-home interference and strain. Job insecurity was revealed to be related to work-related smartphone use and smartphone use to work-home interference. Smartphone use was also found to partially mediate the path from job insecurity to work-home interference. Finally, to prevent negative outcomes of strain, work-home interference and job insecurity, organizations should focus on: providing culture that fits their employees, try to plan availability free time and provide a clear communication.
Keywords: Smartphone use, work-home interference, strain, stress, job insecurity, segmentation preferences
Published in DKUM: 14.01.2019; Views: 1847; Downloads: 137
.pdf Full text (863,72 KB)

5.
Cluster analysis as a tool of guests segmentation by the degree of their demand
Damijan Mumel, Boris Snoj, 2002, original scientific article

Abstract: Authors demonstrate the use of cluster analysis in findin out (ascertaining) the homogenity/heterogenity of guests as to the degree of their demand. The degree of guests' demand is defined according to the importance of perceived service quality components measured by SERVQUAL, which was adopted and adapted, according to the specifics of health spa industry in Slovenia. Goals of the article are: (a) the identification of the profile of importance of general health spa service quality components, and (b) the identification of groups of guests (segments) according to the degree of their demand in the research in 1991 compared with 1999. Cluster analysis serves as useful tool for guest segmentation since it reveals the existence of important differences in the structure of guests in the year 1991 compared with the year 1999. The results serve as a useful database for management in health spas.
Keywords: catering, hotel management, analysis, segmentation, services, quality
Published in DKUM: 04.07.2017; Views: 799; Downloads: 98
.pdf Full text (1,12 MB)
This document has many files! More...

6.
Online speech/music segmentation based on the variance mean of filter bank energy
Marko Kos, Matej Grašič, Zdravko Kačič, 2009, original scientific article

Abstract: This paper presents a novel feature for online speech/music segmentation basedon the variance mean of filter bank energy (VMFBE). The idea that encouraged the feature's construction is energy variation in a narrow frequency sub-band. The energy varies more rapidly, and to a greater extent for speech than for music. Therefore, an energy variance in such a sub-band isgreater for speech than for music. The radio broadcast database and the BNSIbroadcast news database were used for feature discrimination and segmentation ability evaluation. The calculation procedure of the VMFBE feature has 4 out of 6 steps in common with the MFCC feature calculation procedure. Therefore, it is a very convenient speech/music discriminator for use in real-time automatic speech recognition systems based on MFCC features, because valuable processing time can be saved, and computation load is only slightly increased. Analysis of the feature's speech/music discriminative ability shows an average error rate below 10% for radio broadcast material and it outperforms other features used for comparison, by more than 8%. The proposed feature as a stand-alone speech/music discriminator in a segmentation system achieves an overall accuracy of over 94% on radio broadcast material.
Keywords: online speech segmentation, algorithm, speech techniques
Published in DKUM: 26.06.2017; Views: 1337; Downloads: 449
.pdf Full text (1,49 MB)
This document has many files! More...

7.
Detection of planar points for building extraction from LiDAR data based on differential morphological and attribute profiles
Domen Mongus, Niko Lukač, Denis Obrul, Borut Žalik, 2013, published scientific conference contribution

Abstract: This paper considers a new method for building-extraction from LiDAR data. This method uses multi-scale levelling schema or MSLS-segmentation based on differential morphological profiles for removing non-building points from LiDAR data during the data denoising step. A new morphological algorithm is proposed for the detection of flat regions and obtaining a set of building-candidates. This binarisation step is made by using differential attribute profiles based on the sum of the second-order morphological gradients. Any distinction between flat and rough surfaces is achieved by area-opening, as applied within each attribute-zone. Thus, the detection of the flat regions is essentially based on the average gradient contained withina region, whilst avoiding subtractive filtering rule. Finally, the shapes of the flat-regions are considered during the building-recognition step. A binary shape-compactness attribute opening is used for this purpose. The efficiency of the proposed method was demonstrated on three test LiDAR datasets containing buildings of different sizes, shapes, and structures. As shown by the experiments, the average quality of the buildings-extraction was more than 95%, with 96%correctness, and 98%completeness. In terms of quality, this method is comparable with TerraScan R , but both methods significantly differ when comparing correctness and completeness of the results.
Keywords: LiDAR, mathematical morphology, segmentation, DAP, DMP, building extraction
Published in DKUM: 10.07.2015; Views: 1527; Downloads: 423
.pdf Full text (2,85 MB)
This document has many files! More...

8.
Segmenting risks in risk management
Borut Jereb, 2009, original scientific article

Abstract: The paper describes a segmentation of risks to make each risk segment more manageable. The proposed approach is primarily intended to improve the confidentiality of risk simulations. The description of the approach is based on a logistics business process system which requires that its input is represented as a process graph. Each process is defined in terms of input and output; input comprises general input as well as risks; output comprises general output as well as impacts. The model takes into consideration internalas well as external input and output. Parameters can be used to define individual processes. Processes include functions that calculate new values of parameters and output on the bases of given input. Based on given tolerance levels for risks, impacts and process parameters, the model determines whether these levels are acceptable. The model assumes that parameters and functions are non-deterministic, i.e. parameters and functions may change in time. Although the approach is described on a very general level, each segment can be further subdivided into subsegments in order to include more characteristics of observed risks.
Keywords: risk, impact, segmentation, risk management, process parameters, logistics, model, simulation tools, non-deterministic
Published in DKUM: 05.06.2012; Views: 2069; Downloads: 53
URL Link to full text

9.
10.
Search done in 0.2 sec.
Back to top
Logos of partners University of Maribor University of Ljubljana University of Primorska University of Nova Gorica