TU Wien:Computer Vision VU (Sablatnig)/Prüfung 2020-12-18
Zur Navigation springen
Zur Suche springen
Thin Lens Equation[Bearbeiten | Quelltext bearbeiten]
- What is described by the thin lens equation? Describe all variables of the equation
Image Pyramids[Bearbeiten | Quelltext bearbeiten]
- How are gaussian pyramids generated and what is their frequency composition on the different pyramid levels?
- Following the above question, what is the frequency composition of a laplacian pyramid?
Deep Learning[Bearbeiten | Quelltext bearbeiten]
- Explain the concept and application of convolutional neural networks.
- What is the difference between "classical" machine learning pipeline for image recognition and new deep learning approaches?
Stereo[Bearbeiten | Quelltext bearbeiten]
Describe briefly the following 6 terms of stereo vision:
- Disparity
- Baseline
- Epipole
- Epipolar Line
- Epipolar plane
- Essential Matrix
Multiview Reconstruction[Bearbeiten | Quelltext bearbeiten]
- Given many points in correspondence across several images, what is the goal of Structure-from-motion?
- What is the input and output of bundle adjustment? What is minimized during the bundle adjustment process?
Machine Learning[Bearbeiten | Quelltext bearbeiten]
- What is regularization in machine learning and why is it done?
- Describe the difference between supervised and unsupervised machine learning
Image Features[Bearbeiten | Quelltext bearbeiten]
- How is rotation invariance achieved by SIFT?
- What is scale- and rotation invariance in the context of local image descriptors and interest point detectors? Give an example.
Image acquisition[Bearbeiten | Quelltext bearbeiten]
- What are the internal camera parameters (sketch), and what influence do they have?
- What are the external camera parameters (sketch), and what influence do they have?
Filtering[Bearbeiten | Quelltext bearbeiten]
True/False questions
- A gaussian filter blurs the image
- Linear filters can be used for sharpening the image
- The Fourier Transform is a local filtering operation
- To compute image gradients two different filters need to be applied to the image.
- Gabor filters can have different orientations
- The difference between the real and imaginary part of a Gabor filter is their wavelength
Depth Perception[Bearbeiten | Quelltext bearbeiten]
- Explain the concept of canonical viewpoints for human object recognition. In this context, what is the frequency hypothesis and the maximal information hypothesis?
- Shortly describe three monocular depth cues that can be used to infer about the depth of objects in an image.