|
|
Detection of Traffic Signs and Lights
Oškera, Jan ; Špaňhel, Jakub (referee) ; Herout, Adam (advisor)
The thesis focuses on modern methods of traffic sign detection and traffic lights detection directly in traffic and with use of back analysis. The main subject is convolutional neural networks (CNN). The solution is using convolutional neural networks of YOLO type. The main goal of this thesis is to achieve the greatest possible optimization of speed and accuracy of models. Examines suitable datasets. A number of datasets are used for training and testing. These are composed of real and synthetic data sets. For training and testing, the data were preprocessed using the Yolo mark tool. The training of the model was carried out at a computer center belonging to the virtual organization MetaCentrum VO. Due to the quantifiable evaluation of the detector quality, a program was created statistically and graphically showing its success with use of ROC curve and evaluation protocol COCO. In this thesis I created a model that achieved a success average rate of up to 81 %. The thesis shows the best choice of threshold across versions, sizes and IoU. Extension for mobile phones in TensorFlow Lite and Flutter have also been created.
|
|
|
Supporting Board Game Nemesis on Android Mobile Phone
Štěpánek, Miroslav ; Švec, Tomáš (referee) ; Smrž, Pavel (advisor)
The aim of this thesis is to create a mobile application for the board game Nemesis designed for the Android system, which will allow the user to find out information about the game components during the game. The solution consists of two main parts the first is a model created with the help of the Tensorflow library, which is responsible for the detection of these components. The second is the application itself, which receives results from the model and displays the resulting information to the user. This makes the game easier for the user and helps to speed it up. The resulting system can be modified so that the application can be used for other games.
|
|
|
Exploitation of Machine Learning for Identification of Feeder Rod Movement
Vele, Patrik ; Vašíček, Zdeněk (referee) ; Šimek, Václav (advisor)
The aim of this diploma thesis is to create a device that uses machine learning methods to recognize the movements of a feeder fishing rod based on data from an inertial measurement unit. The introductory part is devoted to the feeder fishing technique, the selection of important movements and the possibilities of attaching the detection device to the rod. This is followed by the creation of a theoretical basis in the field of machine learning, familiarization with the inertial measurement unit and the issue of classification. The acquired knowledge is used to select appropriate techniques for solving the task of recognizing the movements of the rod. In the practical part, a detection device based on the ESP32 platform is designed and created. This is initially used as a motion sensor, which, in combination with the processing of the measured values, serves as a generator of a training data set. The work continues with the implementation of the convolutional neural network, the learning process on the created dataset and the integration of the most successful model into the detection device. The conclusion is devoted to testing in practice, evaluation and possibilities of future development. The result is a small, battery-powered device that, when attached to any feeder rod, provides highly successful detection of all key movements during the hunt. In addition, thanks to wireless communication via ESP-NOW, it is possible to send the results to various devices.
|
|
|
Conversion of Models between Machine Learning Frameworks for Mobile Platforms
Pavella, Martin ; Zbořil, František (referee) ; Kočí, Radek (advisor)
Machine learning frameworks use various formats to represent and store models of deep neural networks (DNN). One of the most commonly used ones is Open Neural Network Exchange (ONNX). Developing drivers for hardware accelerators on embedded systems is expensive, and ONNX is rarely supported. The necessary software support is typically only implemented for the TensorFlow Lite (TFLite) DNN model format. Currently, the options for conversion of pre-trained ONNX models to TFLite are inadequate and produce suboptimal models. This work focuses on designing and developing a direct converter of ONNX models to TFLite, which produces as optimal models as possible. The resulting program was verified on real models in collaboration with the NXP company. The models produce identical outputs after conversion and their inference speed on target platforms is significantly higher.
|
|
|
Exploitation of Machine Learning for Identification of Feeder Rod Movement
Vele, Patrik ; Vašíček, Zdeněk (referee) ; Šimek, Václav (advisor)
The aim of this diploma thesis is to create a device that uses machine learning methods to recognize the movements of a feeder fishing rod based on data from an inertial measurement unit. The introductory part is devoted to the feeder fishing technique, the selection of important movements and the possibilities of attaching the detection device to the rod. This is followed by the creation of a theoretical basis in the field of machine learning, familiarization with the inertial measurement unit and the issue of classification. The acquired knowledge is used to select appropriate techniques for solving the task of recognizing the movements of the rod. In the practical part, a detection device based on the ESP32 platform is designed and created. This is initially used as a motion sensor, which, in combination with the processing of the measured values, serves as a generator of a training data set. The work continues with the implementation of the convolutional neural network, the learning process on the created dataset and the integration of the most successful model into the detection device. The conclusion is devoted to testing in practice, evaluation and possibilities of future development. The result is a small, battery-powered device that, when attached to any feeder rod, provides highly successful detection of all key movements during the hunt. In addition, thanks to wireless communication via ESP-NOW, it is possible to send the results to various devices.
|
|
|
Supporting Board Game Nemesis on Android Mobile Phone
Štěpánek, Miroslav ; Švec, Tomáš (referee) ; Smrž, Pavel (advisor)
The aim of this thesis is to create a mobile application for the board game Nemesis designed for the Android system, which will allow the user to find out information about the game components during the game. The solution consists of two main parts the first is a model created with the help of the Tensorflow library, which is responsible for the detection of these components. The second is the application itself, which receives results from the model and displays the resulting information to the user. This makes the game easier for the user and helps to speed it up. The resulting system can be modified so that the application can be used for other games.
|
|
|
Detection of Traffic Signs and Lights
Oškera, Jan ; Špaňhel, Jakub (referee) ; Herout, Adam (advisor)
The thesis focuses on modern methods of traffic sign detection and traffic lights detection directly in traffic and with use of back analysis. The main subject is convolutional neural networks (CNN). The solution is using convolutional neural networks of YOLO type. The main goal of this thesis is to achieve the greatest possible optimization of speed and accuracy of models. Examines suitable datasets. A number of datasets are used for training and testing. These are composed of real and synthetic data sets. For training and testing, the data were preprocessed using the Yolo mark tool. The training of the model was carried out at a computer center belonging to the virtual organization MetaCentrum VO. Due to the quantifiable evaluation of the detector quality, a program was created statistically and graphically showing its success with use of ROC curve and evaluation protocol COCO. In this thesis I created a model that achieved a success average rate of up to 81 %. The thesis shows the best choice of threshold across versions, sizes and IoU. Extension for mobile phones in TensorFlow Lite and Flutter have also been created.
|
|
|
Detection Of Collapse By Android Smartphone
Repčík, Tomáš
The bachelor’s study is focused to design and build an Android application for the detection of collapse, which is enhanced by new techniques coming from a sphere of the artificial intelligence modified for smartphones. The application uses accelerometer outputs which are in suspicious moments analysed by the neural network. The artificial intelligence is based on simulated events of collapse and events which resemble a fall of a person. The study describes data collected from 20 people. To provide the best results of training, the most convenient and useful features were selected by multiple approaches. Total accuracy of the collapse detection reached 93 %, with 9 % and 13 % of false positive and false negative detections, respectively.
|
guest ::
