|
|
Prediction of the effect of mutations on protein stability
Rosinská, Monika ; Martínek, Tomáš (referee) ; Musil, Miloš (advisor)
Cílem této práce bylo vytvořit klasifikační model, který dokáže rozlišovat mutace proteinu jako stabilizující či destabilizující. Model byl implementován pomocí konvolučních neuronových sítí s využitím předtrénované sítě ResNet50V2 a datové sady založené na FireProtDB. Model dokáže predikovat vliv mutací na stabilitu proteinu s přesností 0.7 a zároveň poskytuje informaci o míře jistoty dané predikce.
|
|
|
Evolutionary Design of Neural Networks
Kastner, Jan ; Hurta, Martin (referee) ; Sekanina, Lukáš (advisor)
The thesis deals with the implementation of a problem-solving method for the automated design of convolutional neural networks (CNN) architectures. The optimization of two fundamental and often conflicting characteristics, the number of parameters and the quality of CNN classification, is performed using a multi-criteria optimization genetic algorithm (NSGA-II). To encode this problem, the Cartesian genetic programming (CGP) technique is used, which enables the wide range of CNN architectures to be represented, and at the same time, the searched area can be appropriately limited by parameterization. Experiments were performed on the MNIST dataset to understand the effect of population size on the quality of the resulting solution. It is also evident from the results of the experiments that the quality of the architectures found can compete with already established models. This is therefore an alternative approach that does not require human intervention compared to manual design.
|
|
|
Image Inpainting using Deep Learning
Zobaník, Radek ; Kubík, Tibor (referee) ; Šilling, Petr (advisor)
In this thesis, an application was developed for testing and comparing methods for completing missing parts of an image using deep learning, and two methods were trained, pconv with convolutional architecture, and AOT-GAN with GAN architecture. The thesis describes the design of the finished application, its functionality, and important implementation details. A dataset was selected on which the chosen models were optimally trained. Experiments were made on the AOT-GAN model to investigate the impact of the number of AOT blocks in generator on the resulting completed image. All experiments were qualitatively and quantitatively compared. The results showed respectable outcomes when working with natural scenery.
|
|
|
Using unlabeled data for retinal segmentation
Shemshur, Andrii ; Jakubíček, Roman (referee) ; Vičar, Tomáš (advisor)
Tato bakalářská práce se zabývá vývojem a hodnocením pokročilých metod pro segmentaci lékařských snímků v kontextu omezených trénovacích dat. Studie zkoumá techniky učení pod dohledem využívající konvoluční neuronové sítě (CNN), přenosové učení s předtrénovanými modely a strategie učení s částečným dohledem. Jako základní model byl použit model konvoluční neuronové sítě (CNN) s dohledem založený na architektuře U-Net, který dosáhl koeficientu Dice 77,6% a průniku nad sjednocením (IoU) 63,4%. Použití přenosového učení pomocí kodéru ResNet34 předtrénovaného na síti ImageNet vedlo k výraznému zlepšení výkonu s koeficientem Dice 81,9%, IoU 69,3% a přesností 96,7%. Kromě toho byly ke zvýšení výkonu modelu použity strategie učení s částečným dohledem, včetně pseudoznačení a předtrénování denoizace. Přístup pseudoznačení přinesl koeficient Dice 81,7% a IoU 69,1%, čímž prokázal účinnost využití neoznačených dat. Přístup před tréninkem denoizace prokázal robustní výkonnost a dosáhl koeficientu Dice 80,3% a IoU 67,0%, a to i v přítomnosti zašuměných a neoznačených dat. Tyto výsledky podtrhují potenciál transferového učení a poloprovozních metod pro zvýšení přesnosti segmentace při analýze lékařských snímků. Poskytují solidní základ pro budoucí výzkum v této oblasti.
|
|
|
Figure trracking
Berka, Jiří Michael ; Řičánek, Dominik (referee) ; Richter, Miloslav (advisor)
This bachelor's thesis focuses on the development of software for tracking an exercising person and evaluating deviations from the correct execution of movement tasks. Initially, the thesis theoretically examines the fundamentals of digital image processing, the use of various types of cameras, and computer vision technologies. Subsequently, methods of artificial intelligence and deep learning that enable motion detection and analysis are described. The main part of the thesis focuses on the implementation of the OpenPose system for real-time pose estimation. Technical challenges are discussed, and solutions are proposed to achieve the accuracy and reliability of the system. The practical part includes testing the software on real videos and evaluating its performance. The results show that the developed software can effectively help in correcting movements and preventing injuries in various applications.
|
|
|
Signature verification using neural network-based algorithms
Čírtek, Petr ; Kiac, Martin (referee) ; Myška, Vojtěch (advisor)
Signature is one of the most used biometrics in banking and contracting therefore is important to verificate signature authenticity. Verification can be done with the help of a forensic specialist or, thanks to the rise of advanced technology, with the help of a computing technology. The purpose of this thesis is to develop methods for signature verification using neural networks for Czech type of signature and to find out if adding manual extracted features to convolutional analysis could improve these methods. Neural networks seek to replicate the functioning of human brain, consisting of input neurons, several hidden layers and output neurons. Neural networks are one of the most popular artificial intelligence technologies for image analysis and classification. The proposed methods in this thesis work on the principles of convolutional networks. The first proposed method consist of three convolutional layers which extract important features from image of signature and pass them to fully connected classifier layer. This determines whether the signature is genuine or forgery. Also for this method there were created two functions which can interpret it's decision-making. The second method, siamese neural network, unlike the first, does not work with signatures independently, but uses a reference signature image to determine authenticity. The basis of this method is to extract features with convolutional analysis from both the reference signature and the signature to be authenticated. These features are then concatenated and passed to the clasificator. A Czech dataset was created to train models that would verify the Czech type of signatures. From the experiments, it was found that the addition of manualy extracted features has the potential to improve the prediction accuracy of methods based on convolutional image analysis. 3 models were trained, which can verify the Czech type of signatures with an accuracy higher than 80 \%, namely: the model of the convolutional neural network method with discrete wavelet transformation feature, which was trained on the Czech dataset, the model of the same method trained on the CEDAR dataset with number of strokes as added feature and a siamese convolutional neural network method model trained on the Czech dataset of signatures with the tri-surface feature.
|
|
|
Automation of video extensometers using artificial intelligence
Leinweber, Vít ; Adamec, Tomáš (referee) ; Ščerba, Bořek (advisor)
This thesis deals with the automation of the use of video extensometers using artificial intelligence methods, specifically the implementation of a suitable algorithm for the recognition of the type of sample to be measured and the placement of a suitable software tool in the correct position on the sample in real-time. The theoretical part of the thesis describes digital image correlation, computer vision with a focus on object recognition, and machine learning with a focus on deep convolutional neural networks and their architectures for object detection. Based on the findings from the theoretical part of the work, the YOLOv8 algorithm is chosen as the most suitable algorithm for classifying the type of sample and determining its location in real-time. In the practical part of the thesis, a dataset containing three types of samples is created using the database of video extensometer manufacturer. The dataset is extended with images containing samples to be detected and augmented. The selected algorithm is trained, optimized, and tested using the created dataset. A genetic algorithm and a random search of the hyperparameter space are used in the optimization process. The best trained models of the YOLOv8 algorithm are compared with each other on two test sets and the best one is selected. Furthermore, the work with this algorithm as a detector integrated into the corresponding software for working with video extensometers is described. Finally, a method for correcting the effect of potential misalignment of samples in the images entering the detector is proposed.
|
|
|
Automatická kontrola dopravního značení
Čechmánek, Roman ; Klíma, Ondřej (referee) ; Musil, Petr (advisor)
The aim of this work is to create a cost-effective tool capable that would be able to automate the process of traffic sign control. This includes working with records of drives on land communications, created using inexpensive recording devices such as GoPro action cameras or certain dashcams. The control is based on the system localized traffic signs and historical traffic sign mapping data. The result of the work is a system whose input consists of drive records and historical data, and whose output is two files containing information about the inspection results. The first of these is a GEOJSON file, suitable for further processing of the collected data, and an HTML file that provides a simple user interface visualizing the inspection results on an interactive web map.
|
|
|
Evolutionary Design of Non-Linear Functions for Convolutional Neural Networks
Hladiš, Martin ; Mrázek, Vojtěch (referee) ; Sekanina, Lukáš (advisor)
The aim of this thesis is to design and implement a program for automated design of nonlinear activation functions for convolutional neural networks (CNN) using evolutionary algorithms. The use of automated design provides an independent view to systematically explore a wide range of activation functions and identify the best ones. The method for automatic design chosen in this thesis is a form of evolutionary algorithms referred to as Cartesian genetic programming, which uses a graph representation to encode the solution. This technique allows for the definition of a set of mathematical primitives that define the search space, and thus simply parameterize the design. The implemented approach has been tested on several different architectures and datasets (LeNet-5 \& MNIST, ResNet-10 \& FashionMNIST, WRN-40-4 \& CIFAR-10). Experiments have shown that the approach can find activation functions that statistically improve the accuracy of the architecture over the commonly used ReLU function.
|
|
|
Retinal Images Generation with a Limited Amount of Training Data
Senichak, Yahor ; Semerád, Lukáš (referee) ; Kavetskyi, Andrii (advisor)
The purpose of this study is to explore the progress and application of computer vision and generative adversarial networks (GANs3.1) in the diagnosis and study of fundus diseases. Particular attention is paid to the latest advances in the field of medical data synthesis and the development of our own algorithm. Recent advances in the deep learning architecture U-GAT-IT [22], which includes two pairs of deep neural networks (two generators and two discriminators), have been implemented. This implementation was trained for approximately 300,000 iterations, during which positive results were obtained. The dynamics of the training process were recorded and tests were performed to demonstrate the ability to generate high-quality synthetic images of the ocular background independent of the input data
|
guest ::
