|
|
Object Detection on GPU
Jurák, Martin ; Hradiš, Michal (referee) ; Juránek, Roman (advisor)
This thesis is focused on the acceleration of Random Forest object detection in an image. Random Forest detector is an ensemble of independently evaluated random decision trees. This feature can be used to acceleration on graphics unit. Development and increasing performance of graphics processing units allow the use of GPU for general-purpose computing (GPGPU). The goal of this thesis is describe how to implement Random Forest method on GPU with OpenCL standard.
|
|
|
Application of AdaBoost
Wrhel, Vladimír ; Šilhavá, Jana (referee) ; Hradiš, Michal (advisor)
Basics of classification and pattern recognitions will be mentioned in this work. We will focus mainly on AdaBoost algorithm, which serves to create a strong classifier function by some weak classifiers. We shall get acquainted with some modifications of AdaBoost. These modifications improve some of AdaBoost attributes. We shall also look into weak classifiers and features applicable to them. We shall especially look into the Haar- likes features. We shall discus possibilities of using the mentioned algorithms and features in facial expression recognition. We shall describe the situation between facial expression databases. We shall draw out a possible implementation of application of facial expression recognition.
|
|
|
Hand-Held 3D Scanner
Kukučka, Marek ; Španěl, Michal (referee) ; Hradiš, Michal (advisor)
The goal of this thesis is to implement a method of 3D reconstruction from pairs of images taken using a device that consists of two cameras. The method of stereo vision was chosen and the result is a sparse reconstruction of the scanned object. First, the cameras are calibrated and the acquired images are modified. In the next section, we look for corresponding key points in a pair of images. After obtaining the corresponding points, we an then perform their triangulation. For reconstruction from more then two consecutive images, we use recalculation of projection matrix. In this thesis an experiment is performed, with aim to test whether the reconstructed object corresponds in its dimensions to the real world.
|
|
|
License Plate Detection and Recognition for Traffic Analysis
Černá, Tereza ; Hradiš, Michal (referee) ; Herout, Adam (advisor)
This thesis describes the design and development of a system for detection and recognition of license plates. The work is divided into three basic parts: licence plates detection, finding of character positions and optical character recognition. To fullfill the goal of this work, a new dataset was taken. It contains 2814 license plates used for training classifiers and 2620 plates to evaluate the success rate of the system. Cascade Classifier was used to train detector of licence plates, which has success rate up to 97.8 %. After that, pozitions of individual characters were searched in detected pozitions of licence plates. If there was no character found, detected pozition was not the licence plate. Success rate of licence plates detection with all the characters found is up to 88.5 %. Character recognition is performed by SVM classifier. The system detects successfully with no errors up to 97.7 % of all licence plates.
|
|
|
Neural Networks for Autonomous Car Driving
Dopita, Marek ; Hradiš, Michal (referee) ; Smrž, Pavel (advisor)
In this work, the principles of neural networks are introduced with a focus on autonomous vehicles. Based on this information, a proposal for the implementation of a system is created, which allows to drive a car without a driver. It builds on tools that allow easy creation and testing of autonomous vehicles. It is CARLA simulator and ranking.The proposal divides vehicle routes into three different situations. Each situation requires the use of different sensors, so a specific autonomous agent is created that is able to recognize the situation and switch between different neural network designs. Each such network is specific in its inputs and is taught in a specific situation.Programs are created that are able to easily collect a data set using the CARLA Leaderboard. Then, a way is introduced to how the collected data can be divided into categories so that each category can be used to learn its neural network.
|
|
|
WaldBoost on GPU
Polok, Lukáš ; Mikolov, Tomáš (referee) ; Hradiš, Michal (advisor)
Image recognition and machine vision in general is quickly evolving field, due boom of cheap and powerful computation technologies. Image recognition has many different applications in wide spectrum of industries, ranging from communications trough security to entertainment. Algorithms for image recognition are still evolving and are often quite computationaly demanding. That is why some of authors deal with implementing the algorithms on specialized hardware accelerators. This work describes implementation of image recognition using the WaldBoost algorithm on the graphic accelerator (GPU) platform.
|
|
|
Automatic Image Labelling
Sýkora, Michal ; Beran, Vítězslav (referee) ; Hradiš, Michal (advisor)
This work focuses on automatic classification of images into semantic classes based on their contentc, especially in using SVM classifiers. The main objective of this work is to improve classification accuracy on large datasets. Both linear and nonlinear SVM classifiers are considered. In addition, the possibility of transforming features by Restricted Boltzmann Machines and using linear SVM is explored as well. All these approaches are compared in terms of accuracy, computational demands, resource utilization, and possibilities for future research.
|
|
|
Explainable Convolutional Neural Networks
Kamenický, Daniel ; Herout, Adam (referee) ; Hradiš, Michal (advisor)
The aim of this work was to compare several methods for visualizing the features of each class on the input pixel layer of the CNN. Each method uses a different algorithm, based on gradients, to compute the resulting values. Using the implementation of each method, the resultant values of the methods are obtained by using the equation of energy concentration. The resultant values are presented in tables and graphs from which the success rate of the result of the work can be read. The difference between the methods and comparison of their results can be read from the work. This makes it possible to get an overview of gradient based visualization methods.
|
|
|
Localisation of Mobile Robot in the Environment
Němec, Lukáš ; Hradiš, Michal (referee) ; Veľas, Martin (advisor)
This paper addresses the problem of mobile robot localization based on current 2D and 3D data and previous records. Focusing on practical loop detection in the trajectory of a robot. The objective of this work was to evaluate current methods of image processing and depth data for issues of localization in environment. This work uses Bag of Words for 2D data and environment of point cloud with Viewpoint Feature Histogram for 3D data. Designed system was implemented and evaluated.
|
|
|
Algorithmic Accompaniment Composition
Vinš, Jakub ; Hradiš, Michal (referee) ; Kolář, Martin (advisor)
This thesis deals with problems of computer music, especially with generating accompaniment to an existing song in MIDI format by means of artificial neural networks. Existing methods of algorithmic music composition are presented in the beginning. Followed by problems and their solutions connected with the conversion of MIDI files to matrices, which are suitable as an input for neural network and their inverse transformation. Subsequently are proposed, created, optimized and evaluated models which generate saxophone and piano accompaniment by means of feedforward and recurrent neural network. At the end model generates accompaniment to my own song as a form of a test.
|
guest ::
