|
|
Support of Mapping by Image Processing
Jaroš, Ján ; Herman, David (referee) ; Váňa, Jan (advisor)
This bachelor's thesis deals with methods of detection of selected objects in video and with importing these objects into OpenStreetMap central database based on their geographic location. It focuses mainly on recognition of road signs. First section briefly describes some of the most widely used methods and OpenStreetMap project itself. In the following chapters is given a more detailed overview of used methods of proposed system, its implementation and testing. The conclusion contains evaluation of whole work and the possible improvements are listed here.
|
|
|
Generating training data with neural networks
Ševčík, Pavel ; Kolář, Martin (referee) ; Hradiš, Michal (advisor)
The aim of this thesis was to prepare a training data set for traffic sign detection using generative neural networks. The solution uses a modified U-Net architecture and several experiments with the application of styles using AdaIN layers as in the StyleGAN model have been conducted. By extending the real GTSDB data set with the generated images, mean average precision of 80.36 % has been achieved, which yields an improvement of 19.27 % compared to the mean average precision of the detection model trained on real data only.
|
|
|
SMART CAR: Traffic Signs Detection
Brzoza, Martin ; Motlíček, Petr (referee) ; Beran, Vítězslav (advisor)
This bachelor's thesis is concerned about known methods of detection traffic signs in video sequence. In introduction are explained and located main benefits in the light of process speed and accuracy of detection. At the next chapters is designed system for traffic signs detection and classification. In the end is this system tested on prepared testing data and results are evaluated. The work is focused on method Template Matching for classification and color appearance model CieCam97 for detecting candidate areas.
|
|
|
Vehicle On-Board Camera Analysis
Kadeřábek, Jan ; Bartl, Vojtěch (referee) ; Špaňhel, Jakub (advisor)
This thesis focuses on analysis of video from vehicle on-board camera. During the process of analysis, probihibitory traffic signs are detected and their specific type is classified. For recognized speed limit signs, their numeric value is extracted. From the processed information, it will try to create a file containing the unique occurrences of traffic signs including their GPS coordinates. For the purpose of detection and recognition of traffic signs, several data sets were created. A~cascade classifier with LBP features is used as a detector. Classification of the type and value of traffic signs is done using the k-Nearest Neigbour method.
|
|
|
Improving Accuracy of Detection and Recognition of Traffic Signs with GANs
Glos, Michal ; Musil, Petr (referee) ; Smrž, Pavel (advisor)
The goal of this thesis was to extend a dataset for traffic sign detection. The solution was based on generative neural networks PatchGAN and Wasserstein GAN of combined DenseNet and U-Net architecture. Those models were designed to synthesize real looking traffic signs from images of their norms. Model for object detection SSD, trained on synthetic data only, achieved mean average precision of 59.6 %, which is an improvement of 9.4 % over the model trained on the original data. SSD model trained on synthetic and original data combined achieved mean average precision of 80.1 %.
|
|
|
Design of traffic sign detector using image processing methods
Šmíd, Josef ; Adámek, Roman (referee) ; Věchet, Stanislav (advisor)
This master thesis deals with the design of a traffic sign detector using the image processing methods. The OpenCV library for working with images in programming language Python is used for this. The first part reports on the using methods. In the next part, these methods were tested on images of traffic signs taken in traffic in different lighting conditions. The results of these tests led to the design of optimal methods and their settings, which were re-verified by verifying on video of driving in traffic. This also revealed the conditions under which they can operate in real-time systems. Finally, an optimization algorithm for compensation of detection errors was proposed from the monitoring of detection waveforms.
|
|
|
Detection and Recognition of Traffic Signs
Vránsky, Radovan ; Beran, Vítězslav (referee) ; Herout, Adam (advisor)
This bachelor thesis is about different methods of detection and recognition of traffic signs in pictures. The introduction several of these methods are described and their use is demonstrated. In the next part of the thesis, the implementation of the detection and recognition of traffic signs with the use of Support Vector Machine is described in detail. It also describes the method of creating of the dataset or different models describing this dataset. In the conclusion the method is evaluated.
|
|
|
Improving Accuracy of Detection and Recognition of Traffic Signs with GANs
Glos, Michal ; Musil, Petr (referee) ; Smrž, Pavel (advisor)
The goal of this thesis was to extend a dataset for traffic sign detection. The solution was based on generative neural networks PatchGAN and Wasserstein GAN of combined DenseNet and U-Net architecture. Those models were designed to synthesize real looking traffic signs from images of their norms. Model for object detection SSD, trained on synthetic data only, achieved mean average precision of 59.6 %, which is an improvement of 9.4 % over the model trained on the original data. SSD model trained on synthetic and original data combined achieved mean average precision of 80.1 %.
|
|
|
Design of traffic sign detector using image processing methods
Šmíd, Josef ; Adámek, Roman (referee) ; Věchet, Stanislav (advisor)
This master thesis deals with the design of a traffic sign detector using the image processing methods. The OpenCV library for working with images in programming language Python is used for this. The first part reports on the using methods. In the next part, these methods were tested on images of traffic signs taken in traffic in different lighting conditions. The results of these tests led to the design of optimal methods and their settings, which were re-verified by verifying on video of driving in traffic. This also revealed the conditions under which they can operate in real-time systems. Finally, an optimization algorithm for compensation of detection errors was proposed from the monitoring of detection waveforms.
|
|
|
Generating training data with neural networks
Ševčík, Pavel ; Kolář, Martin (referee) ; Hradiš, Michal (advisor)
The aim of this thesis was to prepare a training data set for traffic sign detection using generative neural networks. The solution uses a modified U-Net architecture and several experiments with the application of styles using AdaIN layers as in the StyleGAN model have been conducted. By extending the real GTSDB data set with the generated images, mean average precision of 80.36 % has been achieved, which yields an improvement of 19.27 % compared to the mean average precision of the detection model trained on real data only.
|
guest ::
