National Repository of Grey Literature 249 records found  1 - 10nextend  jump to record: Search took 0.00 seconds. 
Deployment of deep learning-based anomaly detection systems: challenges and solutions
Ježek, Štěpán ; Burget, Radim
Visual anomaly detection systems play an important role in various domains, including surveillance, industrial quality control, and medical imaging. However, the deployment of such systems presents significant challenges due to a wide range of possible scene setups with varying number of devices and high computational requirements of deep learning algorithms. This research paper investigates the challenges encountered during the deployment of visual anomaly detection systems for industrial applications and proposes solutions to address them effectively. We present a model use case scenario from real-world manufacturing quality control and propose an efficient distributed system for deployment of the defect detection methods in manufacturing facilities. The proposed solution aims to provide a general framework for deploying visual defect detection algorithms base on deep neural networks and their high computational requirements. Additionally, the paper addresses challenges related the whole process of automated quality control, which can be performed with varying number of camera devices and it mostly requires interaction with other factory services or workers themselves. We believe the presented framework can contribute to more widespread use of deep learning-based defect detection systems, which may provide valuable feedback for further research and development.
Deep Learning for Agar Plate Analysis: Predicting Microbial Cluster Counts
Čičatka, Michal ; Burget, Radim
Manual analysis of agar plates remains a bottleneck in microbiology, hindering automation efforts. This study investigates the feasibility of using machine learning for automated microbial cluster count detection from agar plate images. We employed various methods, including elbow detection (baseline) and supervised learning models (Support Vector Regression, Simple CNN, XGBoost, Random Forest, pre-trained VGG, and pre-trained Inceptionv3). The results demonstrate that machine learning models significantly outperform the baseline, achieving lower prediction errors and higher accuracy in identifying the correct number of clusters. Notably, both pre-trained VGG and InceptionV3 achieved strong performance, highlighting the effectiveness of transfer learning for this task. InceptionV3 exhibited the lowest error rates overall. This study establishes a foundation for developing robust automated systems for quantifying microbial growth, potentially streamlining workflows and improving efficiency in microbiological research and clinical settings.
Enhancing Localization Accuracy in Industrial Wearables with LoRaWAN
Svertoka, Ekaterina ; Martian, Alexandru (referee) ; Digulescu-Popescu,, Angela (referee) ; Lohan, Elena Simona (referee) ; Hošek,, Jiří (referee) ; Burget, Radim (advisor)
Tento výzkum kombinuje teoretické poznatky, simulační studie a praktické experimenty s cílem prozkoumat oblast průmyslových nositelných zařízení se zaměřením na zvýšení přesnosti jejich lokalizace pomocí technologie LoRaWAN. Práce vedla k vytvoření dvou klasifikací funkcí a metrik průmyslových nositelných zařízení a 7 volně přístupných datových sad LoRaWAN v různých prostředích (vnitřních, venkovních a podzemních). Kromě toho práce provádí komplexní posouzení přesnosti lokalizace pomocí více přístupů, analyzuje vliv proměnných z měřicí kampaně a technik zpracování dat. Dále navrhuje úpravy algoritmu k-NN, které spolu s metodami předzpracování vedou ke zvýšení přesnosti o 17,2 % ve srovnání s původním benchmarkem. Navržené algoritmy, ověřené na souborech dat LoRaWAN, nabízejí potenciální využití v různých oblastech. Studie uzavírá validaci lokalizace na bázi LoRaWAN s přesností 2,6 m v interiéru a 4 m v exteriéru, což naznačuje, že ačkoli lokalizace na bázi LoRaWAN není tak přesná jako u předních technologií, lze ji využít v odvětvích, jako je logistika, zemědělství a inteligentní výroba, kde není absolutní přesnost nezbytná.
Personalized Treatment of Respiratory Diseases Using Artificial Intelligence and Interoperability with e-Health Systems
Myška, Vojtěch ; Drotár,, Peter (referee) ; Brezany, Peter (referee) ; Burget, Radim (advisor)
Corticosteroid (CS) treatment in patients with Long COVID aims to prevent the progression from active post-inflammatory changes to fibrosis scarring. However, CS have side effects, which may sometimes be severe. Some patients might not require any treatment as their post-inflammatory changes resolve spontaneously. This dissertation thesis aims to develop an artificial intelligence (AI) based approach that allows personalized treatment of patients with Long COVID and a design of modular architecture allowing seamless interoperability of AI models with the information systems used in healthcare facilities. The first part of the thesis deals with the foundation of the state-of-the-art of using AI algorithms to recommend CS treatment in patients with Long COVID, who have the risk of permanent lung damage. This study examines how various parameters from different examinations influence the accuracy of the AI models. The most effective model achieves an accuracy of 73.68 %, a balanced accuracy of 73.52 %, and an AUC of 0.7469. These results prove that a trained AI model on a correctly chosen set of parameters from various medical examinations is effective and can be used as a decision-support tool for further treatment courses. The second part focuses on developing a modular architecture that allows interoperability between AI models and the information system of health facilities. Its specific implementation for early COVID-19 detection, incorporating DeepCovidXR models, is presented. In the performance test, the average processing time of X-ray images is 11.53 seconds using the CPU and 2.78 seconds with the GPU. Both values meet the maximum permissible analysis time set at 20 seconds. The results presented in both sections have been implemented and are currently used at the Olomouc University Hospital.
A modern approach to measuring antibiotic susceptibility of microbial cultures using machine learning
Lepík, Jakub ; Burget, Radim (referee) ; Čičatka, Michal (advisor)
The bachelor's thesis focuses on antibiotic susceptibility testing (AST), specifically enhancing and automating the assessment of the disk diffusion method using machine learning and object detection architectures. Thanks to the TensorFlow development platform and extensive dataset, on which custom detection models like EfficientDet were trained, processing a wide range of input data is enabled. This brings the possibility of using mobile devices alongside traditional laboratory equipment when evaluating this method. By employing additional image processing techniques and the OpenCV library, a custom algorithm for measuring the size of inhibitory zones was developed, which, along with the detection models, is integrated within the application module developed by Bruker Daltonics GmbH & Co. KG. This module, created using the ASP.NET platform, is a precise and valuable tool for assisting personnel in microbiological laboratories.
Automatic extraction of knowledge from medical reports to minimize the risks of human error
Tománek, Stanislav ; Mezina, Anzhelika (referee) ; Burget, Radim (advisor)
This bachelors thesis focuses on creation of datasets for trainings models for the purpose of summarizing medical reports and text analysis to determine whether a patient is a smoker, has a couch or suffers from pneumonia. Training techniques are introduced from basic training to creating mini LoRA models in a home environment to maintain private data from the reach of third parties.
Forensic method for recognizing the authenticity of artworks using multispectral analysis
Lánský, David ; Mezina, Anzhelika (referee) ; Burget, Radim (advisor)
Detecting forgeries is crucial for protecting the art market and preserving the authenticity of artworks. This thesis focuses on forgery detection using convolutional neural networks (CNNs). The main goal was to develop advanced methods capable of identifying anomalies, and thus potential forgeries, in images with their X-ray photographs. During this research, U-net architectures and binary semantic segmentation techniques were applied, enabling successful anomaly detection. The main contribution of this work is 112 models of four different U-net and U-net++ architectures, which effectively highlight anomalies through the method of binary semantic segmentation. The models were trained on a set of images with their synthetically created X-ray images and artificially generated anomalies. In this way, the models can detect lead spots, nails, layers of hidden paintings, and other defects, while also being able to ignore insignificant elements, such as picture frames and overexposed X-ray images. The testing of the models occurred in two phases. In the first phase, they were evaluated using the IoU metric on a set of 400 synthetically generated data, where in the best cases, they achieved up to 83.5 % IoU. In the second phase, they were evaluated subjectively on images with real X-rays and natural anomalies. This approach combines traditional X-ray techniques with modern computer vision, revealing deviations that might be overlooked during standard visual inspection. By bridging these technologies, this work opens new possibilities for the protection of art collections and provides a solid foundation for further research in the field of art forgery detection using artificial intelligence.
Optimization of control using reinforcement learning on the Robocode platform
Pastušek, Václav ; Myška, Vojtěch (referee) ; Burget, Radim (advisor)
This master's thesis focuses on optimizing the control of a tank robot in the Robocode environment using reinforcement learning. The complexity of this problem falls into the EXPSPACE class, presenting a challenge that cannot be underestimated. The theoretical part of the thesis meticulously examines the Robocode platform, concepts of reinforcement learning, and relevant algorithms, while the practical part focuses on optimizing the agent, implementing reinforcement learning algorithms, and creating a user-friendly interface for easy training and testing of models. A total of 64 models were trained and tested as part of the thesis, with their data and parameters compared and presented in accompanying databases and graphs. The best results in terms of average hits per episode were achieved by models labeled v0.8.0 and v1.0.0. The first model exhibited a certain ability to evade shots, while the second model showed more successful hits.
Simulation and Optimalization of traffic for Smart Cities
Petrák, Tomáš ; Burget, Radim (referee) ; Fujdiak, Radek (advisor)
The thesis is dealing with traffic management using telemetry networks. The problematic of telemetry networks and multiagent systems. A simulation model is proposed in Java which enables configuration simulation and assessment.
Feature extraction from image data
Uher, Václav ; Beneš, Radek (referee) ; Burget, Radim (advisor)
Image processing is one area of signal analysis. This thesis is involved in feature extraction from image data and its implementation using Java programming language. The main contribution of this thesis lies in develop features extractors and their implementation in the program RapidMiner. The result is a robust tool for image analysis. The functionality of each operator is tested on mammogram images. A function model was developed for the removal of artifacts from the mammography images. The success rate of removal is comparable with other similar works. Furthermore, learning algorithms were compared on example detection of ventricle in ultrasound image.

National Repository of Grey Literature : 249 records found   1 - 10nextend  jump to record:
Interested in being notified about new results for this query?
Subscribe to the RSS feed.