|
|
Animation of Avatar Face Based on Human Face Video
Takács, Martin ; Milet, Tomáš (referee) ; Herout, Adam (advisor)
This thesis presents an application for animating 3D avatar based on a single camera or video input of human face in real time. The resulting application consists of three modules -- face tracking, avatar animator, and a server for transferring face data. The face tracking module computes new transforms for the animation from human face and benefits from Mediapipe's Facemesh to estimate the current face geometry. Avatar animator module is a web-based application for rendering and animating 3D avatars through skeletal animations, based on the Three.js library. Both modules make use of the continuous bidirectional communication of websockets through the single server. Performance of the face tracking module depends on the camera and device on which it is running, but regular web camera is usually enough for speed of 30+ FPS and animation runs at the speed of 60+ FPS with multiple avatars. Main contributions of this project are (a) Calculating transforms from human face is suitable for the skeletal animations, which are usually easier to create and more accessible than blend shapes. (b) Using the face tracking and the avatar animator as independent modules causes that from the single camera/video input it is possible to animate multiple avatars on different devices.
|
|
|
Face Detection in Video
Kolman, Aleš ; Řezníček, Ivo (referee) ; Polok, Lukáš (advisor)
The project is focused on face detection in video. Firstly, it contains a summary of basic color models. Secondly, you can find the description and comparison of the basic methods for detection of human skin with a practical example of implementation of parametric detector. Thirdly, a theoretical basis for face detection and face tracking in a video containing a list of basic concepts and methods of this issue follows. Greater emphasis is placed on the description of machine learning algorithm AdaBoost and description of the possible application of the Kalman filter for the purpose of face tracking. Design, implementation and testing of library accomplished within the master thesis are listed in the final part of this thesis.
|
|
|
Real-Time Face Tracking
Ermak, Aleksei ; Špaňhel, Jakub (referee) ; Hradiš, Michal (advisor)
This bachelor thesis focuses on the issue of face tracking in real time. In the beginning, this work describes the existing methods of object tracking and face detection. The following part of this thesis concentrates on the design, implementation and testing of the convolutional neural network, which was proved as the effective solution for the face tracking issue. In addition to this, the implemented network is compared to those existing methods. The last part of the thesis describes the optimization of the designed network using OpenVINO toolkit provided by Intel.
|
|
|
Font size adjustment based on distance detection
Brunclík, Robert ; Koťová, Markéta (referee) ; Janoušek, Oto (advisor)
The thesis deals with automatic control the font size by the distance from the reader. It includes theoretical acquaintance with the face detection and subsequent tracking of the detected area during the scene. Furthermore, there is a comparison of the tracking algorithms. Then the calculation of distance is decribed. It is based on the user’s calibration and based on the outcome occurs the font size is automatically corrected. There is also a description of a separate application Automatical controller of the text size, with the recommended settings of the program.
|
|
|
Animation of Avatar Face Based on Human Face Video
Takács, Martin ; Milet, Tomáš (referee) ; Herout, Adam (advisor)
This thesis presents an application for animating 3D avatar based on a single camera or video input of human face in real time. The resulting application consists of three modules -- face tracking, avatar animator, and a server for transferring face data. The face tracking module computes new transforms for the animation from human face and benefits from Mediapipe's Facemesh to estimate the current face geometry. Avatar animator module is a web-based application for rendering and animating 3D avatars through skeletal animations, based on the Three.js library. Both modules make use of the continuous bidirectional communication of websockets through the single server. Performance of the face tracking module depends on the camera and device on which it is running, but regular web camera is usually enough for speed of 30+ FPS and animation runs at the speed of 60+ FPS with multiple avatars. Main contributions of this project are (a) Calculating transforms from human face is suitable for the skeletal animations, which are usually easier to create and more accessible than blend shapes. (b) Using the face tracking and the avatar animator as independent modules causes that from the single camera/video input it is possible to animate multiple avatars on different devices.
|
|
|
Real-Time Face Tracking
Ermak, Aleksei ; Špaňhel, Jakub (referee) ; Hradiš, Michal (advisor)
This bachelor thesis focuses on the issue of face tracking in real time. In the beginning, this work describes the existing methods of object tracking and face detection. The following part of this thesis concentrates on the design, implementation and testing of the convolutional neural network, which was proved as the effective solution for the face tracking issue. In addition to this, the implemented network is compared to those existing methods. The last part of the thesis describes the optimization of the designed network using OpenVINO toolkit provided by Intel.
|
|
|
Font size adjustment based on distance detection
Brunclík, Robert ; Koťová, Markéta (referee) ; Janoušek, Oto (advisor)
The thesis deals with automatic control the font size by the distance from the reader. It includes theoretical acquaintance with the face detection and subsequent tracking of the detected area during the scene. Furthermore, there is a comparison of the tracking algorithms. Then the calculation of distance is decribed. It is based on the user’s calibration and based on the outcome occurs the font size is automatically corrected. There is also a description of a separate application Automatical controller of the text size, with the recommended settings of the program.
|
|
|
Face Detection in Video
Kolman, Aleš ; Řezníček, Ivo (referee) ; Polok, Lukáš (advisor)
The project is focused on face detection in video. Firstly, it contains a summary of basic color models. Secondly, you can find the description and comparison of the basic methods for detection of human skin with a practical example of implementation of parametric detector. Thirdly, a theoretical basis for face detection and face tracking in a video containing a list of basic concepts and methods of this issue follows. Greater emphasis is placed on the description of machine learning algorithm AdaBoost and description of the possible application of the Kalman filter for the purpose of face tracking. Design, implementation and testing of library accomplished within the master thesis are listed in the final part of this thesis.
|
guest ::
