|
|
Stereo Reconstruction with Deep Neural Networks
Letanec, Richard ; Herout, Adam (oponent) ; Španěl, Michal (vedoucí práce)
The aim of this thesis is to design and train a neural network model capable of estimating a disparity map from a pair of images. It will then be possible to create a depth map and point cloud from the estimated disparity map. Such a process is called stereo reconstruction. Solving this task consists of two steps -- choosing a suitable dataset and choosing a suitable neural network architecture. In my work, I compared two neural network architectures that I trained on the DrivingStereo dataset, consisting of paired images photographed from the roof of a car, and retrained and evaluated on the KITTI 2015 dataset, consisting of images of the same type. As the first neural network architecture, I chose ES-Net, which uses an approach based on a sequence of residual blocks and convolutional layers. As the second architecture, I chose CREStereo, which uses an iterative approach based on recurrent layers to predict the disparity map. In all benchmark tests, the CREStereo architecture achieves better accuracy.
|
|
|
Stereo Based 3D Face Reconstruction
Falešník, Milan ; Svoboda, Pavel (oponent) ; Šolony, Marek (vedoucí práce)
This thesis presents a system for reconstruction of three-dimensional model of human face by using pair of cameras. To solve this problem, depth map calculation is used and then the depth map is transformed into three-dimensional space. It uses Viola-Jones detector for detecting the face. Uses libraries OpenCV and partially PCL.
|
|
|
Stereo Based 3D Face Reconstruction
Falešník, Milan ; Svoboda, Pavel (oponent) ; Šolony, Marek (vedoucí práce)
This thesis presents a system for reconstruction of three-dimensional model of human face by using pair of cameras. To solve this problem, depth map calculation is used and then the depth map is transformed into three-dimensional space. It uses Viola-Jones detector for detecting the face. Uses libraries OpenCV and partially PCL.
|
|
|
Camera Calibration by Registration Stereo Reconstruction to 3D Model
Klečka, J.
Paper aims at unusual way to camera calibration. The main idea is that by registration of uncalibrated stereo reconstruction to 3D model of the same scene is eliminated ambiguity of the reconstruction. The reason for this is that exact metric scene reconstruction from image pair can be understate as information equivalent to calibration of the source camera pair. Described principles were verified by experiment on real data and results are presented at the end of the paper.
|
host ::
RSS.