Martin Šinko

ORCID: 0000-0002-9738-376X
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Image and Object Detection Techniques
  • Solar Radiation and Photovoltaics
  • Image Enhancement Techniques
  • Robotics and Sensor-Based Localization
  • Hand Gesture Recognition Systems
  • Neural Networks and Applications
  • Image Retrieval and Classification Techniques
  • Image Processing and 3D Reconstruction
  • Mobile and Web Applications
  • Advanced Malware Detection Techniques
  • 3D Surveying and Cultural Heritage
  • Context-Aware Activity Recognition Systems
  • Advanced Neural Network Applications
  • Interactive and Immersive Displays
  • IoT-based Smart Home Systems
  • Water Quality Monitoring Technologies
  • Photovoltaic System Optimization Techniques
  • Advanced Image Fusion Techniques
  • Medical Image Segmentation Techniques
  • Video Surveillance and Tracking Methods
  • Remote Sensing and LiDAR Applications
  • IoT Networks and Protocols
  • Fire Detection and Safety Systems
  • High-Velocity Impact and Material Behavior
  • 3D Shape Modeling and Analysis

University of Žilina
2018-2020

In this article, we are combining an advanced implementation of the popular ICP algorithm using transformation 3D invariant properties based on scale-invariant feature transform to register free-form closed surfaces (3D model human skull). Unlike point and surface registers, our method better captures bulk nature data such as bone thickness. The proposed is divided into three main steps: function extraction, comparison Euclidean metric distance gross alignment enhancement. input system...

10.1109/elektro.2018.8398245 article EN 2020 ELEKTRO 2018-05-01

This paper proposes a 3D surface registration algorithm based on the iterated closest point (ICP). The proposed uses Scale-Invariant Feature Transform (SIFT) functions for initial alignment in combination with K-Nearst Neighbor (KNN) function comparison and Iterative Closest Point (ICP) weighted performing accurate registration. First, area properties are used corresponding cloud areas. Second, files associated regions classified to calculate transformation matrix. Based this combination,...

10.1109/tsp.2019.8769057 article EN 2019-07-01

In this paper, the comparison between deep learning methods and feature extraction algorithms is presented. The principle of Grey-Level Co-occurrence Matrix (GLCM) its modifications are used for our research. main idea was to design a method description combined features textures. texture classification process carried out with robust support vector machine classifier (SVM). We compare these proposed Convolutional Neural Networks (CNN). This network contains 25 layers. Finally, all...

10.1109/iwssip48289.2020.9145263 article EN 2020-07-01

In this paper a comparison between three feature extraction methods (Fourier Transform, Radon Canny Edge Filter) and Convolutional Neural Network is presented. These are tested on set of depth maps. The Microsoft Kinect camera used for capturing the images. For image classification Support Vector Machine with Radial Basis Function kernel was used. experimental results from each method stored in confusion matrix. Each row matrix represents actual class data column predicted class. quality...

10.23919/ntsp.2018.8524109 article EN 2018-10-01

Weather prediction is a crucial element for power management in photovoltaic plants (PVPP). In this paper, we propose novel system collecting essential data used local short-term weather prediction. Image consists of all-sky ground-based images obtained by an camera with fish-eye lens. Our proposed station collects meteorological into database. The include air temperature, humidity, wind speed, relative pressure, and spectrum solar radiation. First, the whole setup obtaining described,...

10.1016/j.trpro.2019.07.214 article EN Transportation research procedia 2019-01-01

In this paper, a comparison between feature extraction methods (Radon Cosine Method, Canny Contour Fourier Transform, SIFT descriptor, and Hough Lines Method) Convolutional Neural Networks (proposed CNN pre-trained AlexNet) is presented. For the evaluation of these methods, depth maps were used. The tested data obtained by Microsoft Kinect camera (IR sensor). vectors classified Support Vector Machine (SVM). confusion matrix for experimental results was row represents target class column...

10.3849/aimt.01326 article EN cc-by Advances in Military Technology 2020-07-31

This paper addresses the challenge of developing robust object detection systems in context Valve’s Counter-Strike by in-troducing a novel, high-quality dataset generated using complex image generator built within Unity game engine. mimics original game’s environment and character interactions, capturing complexity in-game scenarios. The pro-vides valuable resource for training models like YOLOv9 algorithm, which we employ to develop an system that achieves high precision recall, turn...

10.18690/um.feri.6.2024.6 article EN 2024-10-30

In this paper an artificial neural networks for windows operation system are presented. Artificial essential part of the machine learning algorithms. As they more and used in commercial applications, it is desirable to introduce them educational process. To work with on Windows operating system, Anaconda environment provides simple free charge solution. After introduction topic, basic theory about continues. Next installation description environment. Subsequent chapter defines MNIST example...

10.1109/iceta48886.2019.9040142 article EN 2022 20th International Conference on Emerging eLearning Technologies and Applications (ICETA) 2019-11-01

the comparison of convolutional neural network in Python environment is presented this paper. The Anaconda platform provides free and easy to use tools for scripting language. After introduction environment, experiment described. First used architectures are shown. Used databases defined later. Finally, results presented. testing was done determine computational time needed these architectures.

10.1109/elektro49696.2020.9130321 article EN 2020 ELEKTRO 2020-05-01

This paper proposes about the compact Internet of Things (IoT) module for local weather monitoring. IoT was designed and developed to collect meteorological data purposes power output prediction from photovoltaic panels environmental evaluation air quality in inhabited areas. main component is Arduino board with connected sensors WiFi ESP8266.

10.1109/elektro49696.2020.9130342 article EN 2020 ELEKTRO 2020-05-01

Precise cloud detection and classification is a crucial element for various meteorological applications, such as solar irradiance forecast or cover estimation. In this paper, we present an automatic method cloud-type from ground-based images using improved pretrained deep neural network. the past, there were used algorithms mainly low-level features. Our approach to use artificial network (ANN), namely convolutional (CNN) because these type of networks both types image features: first couple...

10.1109/iceta48886.2019.9040149 article EN 2022 20th International Conference on Emerging eLearning Technologies and Applications (ICETA) 2019-11-01

In this paper the optimization tool of website for smart phone is presented. Internet a part everyday life in last years. People visit websites with different devices. Many methods serve to properly display and optimize web content. These are constantly being improved provide convenient viewing We've created our own app based on designed content correctly. The contains an algorithm identifying main page then extracting text images from it. goal remove redundant information webpage create new...

10.1109/elektro49696.2020.9130244 article EN 2020 ELEKTRO 2020-05-01
Coming Soon ...