|
|
3D Surface Reconstruction From Video Sequences
Dominec, Adam ; Mach, Lukáš (advisor) ; Sedláček, David (referee)
Abstract. This work describes a method for dense scene reconstruction from video, assuming both the external and internal calibration of camera in each frame is known. The method is modular; in the cases of the well studied subproblems, a description of the corresponding algorithms is provided, and where necessary, we present our novel techniques. The method reconstructs the scene in an iterative manner and is highly adaptable to required precision and resolution of the output. This work is accompanied by a complete open- source implementation of the method described.
|
|
|
Crowd Behavior Anomaly Detection in Drone Videodata
Bažout, David ; Herout, Adam (referee) ; Beran, Vítězslav (advisor)
There have been lots of new drone applications in recent years. Drones are also often used in the field of national security forces. The aim of this work is to design and implement a tool intended for crowd behavior analysis in drone videodata. This tool ensures identification of suspicious behavior of persons and facilitates its localization. The main benefits include the design of a suitable video stabilization algorithm to stabilize small jitters, as well as trace back of the lost scene. Furthermore, two anomaly detectors were proposed, differing in the method of feature vector extraction and background modeling. Compared to the state of the art approaches, they achieved comparable results, but at the same time they brought the possibility of online data processing.
|
|
|
Image Registration from Static UAV Platform for Ground Objects Localization
Kučera, Adam ; Luža, Radim (referee) ; Rozman, Jaroslav (advisor)
This paper describes development of new robust method for video registration into shared space. Then it is possible to georegister this video to satellite image using single arbitrary frame. Developed high-level method is based on state-of-the-art low-level image processing algorithms. It is robust to huge and/or instant changes in lighting conditions of the scene and changes in geometry of the view. Global error problem is converted to shortest path optimization problem. Local error is minimized via fusion of two approaches to video stabilization.
|
|
| |
|
|
Deep Neural Networks for Classifying Objects in an Image
Mlynarič, Tomáš ; Zemčík, Pavel (referee) ; Hradiš, Michal (advisor)
This paper deals with classifying objects using deep neural networks. Whole scene segmentation was used as main algorithm for the classification purpose which works with video sequences and obtains information between two video frames. Optical flow was used for getting information from the video frames, based on which features maps of a~neural network are warped. Two neural network architectures were adjusted to work with videos and experimented with. Results of the experiments show, that using videos for image segmentation improves accuracy (IoU) compared to the same architecture working with images.
|
|
|
3D Surface Reconstruction From Video Sequences
Dominec, Adam ; Mach, Lukáš (advisor) ; Sedláček, David (referee)
Abstract. This work describes a method for dense scene reconstruction from video, assuming both the external and internal calibration of camera in each frame is known. The method is modular; in the cases of the well studied subproblems, a description of the corresponding algorithms is provided, and where necessary, we present our novel techniques. The method reconstructs the scene in an iterative manner and is highly adaptable to required precision and resolution of the output. This work is accompanied by a complete open- source implementation of the method described.
|
|
|
Crowd Counting in Video
Kuřátko, Jiří ; Sochor, Jakub (referee) ; Herout, Adam (advisor)
This master's thesis prepared the programme which is able to follow the trajectories of the movement of people and based on this to create various statistics. In practice it is an effective marketing tool which can be used for instance for customer flow analyses, optimal evaluation of opening hours, visitor traffic analyses and for a lot of other benefits. Histograms of oriented gradients, SVM classificator and optical flow monitoring were used to solve this problem. The method of multiple hypothesis tracking was selected for the association data. The system's quality was evaluated from the video footage of the street with the large concentration of pedestrians and from the school's camera system, where the movement in the corridor was monitored and the number of people counted.
|
|
|
Tracking and Recognition of People in Video
Šajboch, Antonín ; Hradiš, Michal (referee) ; Smrž, Pavel (advisor)
The master's thesis deals with detecting and tracking people in the video. To get optimal recognition was used convolution neural network, which extracts vector features from the enclosed frame the face. The extracted vector is further classified. Recognition process must take place in a real time and also with respect are selected optimal methods. There is a new dataset faces, which was obtained from a video record at the faculty area. Videos and dataset were used for experiments to verify the accuracy of the created system. The recognition accuracy is about 85% . The proposed system can be used, for example, to register people, counting passages or to report the occurrence of an unknown person in a building.
|
|
|
Fast Re-Calibration of PTZ Camera for Traffic Analysis
Dřevo, Aleš ; Juránek, Roman (referee) ; Sochor, Jakub (advisor)
This thesis deals with problematics of PTZ-camera re-calibration during movement. The objective of this work is to keep the camera in calibration mode from default status when the known positions of Vanishing Points are in the image. With their use during movement, which is changing with motion of the camera, their positions are kept with help of two implemented methods. The first method is based on the principle of homography, the second on the principle of cross ratio. The results show that both of these methods work especially for keeping the positions of First Vanishing Points. In the case of the Second Vanishing Points there appear various problems and the results are often quite inaccurate.
|
|
|
Detection of parts of human body in an image
Křivánek, Filip ; Přinosil, Jiří (referee) ; Šmirg, Ondřej (advisor)
This thesis deals with algorithms for human figure detection in video. It describes, practically tests and evaluates algorithms called Histogram of Oriented Gradients and Haar Cascade Classifier. Testing program was written in C++ language with use of OpenCV library. The aim is to find the approximate location of individual body parts in the area of the detected human figure. The detection was tested on a video sequence and various image processing methods were selected. Techniques such as thresholding, background subtraction, finding the largest of contours or optical flow were employed. The result of thesis is to find points which represent concrete body part.
|
guest ::
