Computer Vision at Universität Stuttgart | Flashcards & Summaries

Suggested languages for you:

# Lernmaterialien für Computer Vision an der Universität Stuttgart

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen Computer Vision Kurs an der Universität Stuttgart zu.

TESTE DEIN WISSEN

Lösung anzeigen
TESTE DEIN WISSEN

• no assumption about prior needed
• works for non-parametric distributions (multi-modal..)

• requires lots of data
• difficult to determine expedient kernel size
Lösung ausblenden
TESTE DEIN WISSEN

Whats second-order statistics?

Lösung anzeigen
TESTE DEIN WISSEN

incorporating the neighbourhood context

Lösung ausblenden
TESTE DEIN WISSEN

List some major application of SIFT:

Lösung anzeigen
TESTE DEIN WISSEN

1. estimation of global relations between images (stereo geometry, homographics..)

2.) tracking

3.) structur from motion

4.) object detection and pattern recognition

Lösung ausblenden
TESTE DEIN WISSEN

What are the main benefits of Gaussian pyramids?

Lösung anzeigen
TESTE DEIN WISSEN

1.) Solving problems in a coarse to fine manner accelerates most algorithms.

2.) Much more memory efficient as scale-spaces due to the downsampling step

Lösung ausblenden
TESTE DEIN WISSEN

Whats the kurtosis and skewness of a Gaussian?

Lösung anzeigen
TESTE DEIN WISSEN

both 0

Lösung ausblenden
TESTE DEIN WISSEN

Are voxels usually cubic?

Lösung anzeigen
TESTE DEIN WISSEN

No, the voxel dimensions usually differ in different directions. However, pixels are mostly squared.

Lösung ausblenden
TESTE DEIN WISSEN

Is 1 byte sufficent for the co-domain discretisation?

Lösung anzeigen
TESTE DEIN WISSEN

Yes. As a human can only distinguish only around 40 greyscales, 256 are more than sufficient.

Lösung ausblenden
TESTE DEIN WISSEN

What are examples of vector valued images?

Lösung anzeigen
TESTE DEIN WISSEN

1) Colour images RGB -> 3 dimensions

2) Multispectral images

Fun fact: humans can distinguish 2 million colours

Lösung ausblenden
TESTE DEIN WISSEN

Whats quantisation?

Lösung anzeigen
TESTE DEIN WISSEN

Discretisation of the co-domain (wertebereich) --> value discretisation

Lösung ausblenden
TESTE DEIN WISSEN

What are main benefits of preprocessing?

Lösung anzeigen
TESTE DEIN WISSEN

1.) Filtering: removal of noise and artifacts

2.) Simplification: makes it easier to interpret the content

3.) Scaling

4.) Rotating

..

Lösung ausblenden
TESTE DEIN WISSEN

What is regarded as concentration in a diffusion process in computer vision?

Lösung anzeigen
TESTE DEIN WISSEN

grey values

Lösung ausblenden
TESTE DEIN WISSEN

Which classes of invariants do you know?

Lösung anzeigen
TESTE DEIN WISSEN

geometric (e.g. translation, rotation, scaling)

Lösung ausblenden
• 109343 Karteikarten
• 2146 Studierende
• 80 Lernmaterialien

## Beispielhafte Karteikarten für deinen Computer Vision Kurs an der Universität Stuttgart - von Kommilitonen auf StudySmarter erstellt!

Q:

A:

• no assumption about prior needed
• works for non-parametric distributions (multi-modal..)

• requires lots of data
• difficult to determine expedient kernel size
Q:

Whats second-order statistics?

A:

incorporating the neighbourhood context

Q:

List some major application of SIFT:

A:

1. estimation of global relations between images (stereo geometry, homographics..)

2.) tracking

3.) structur from motion

4.) object detection and pattern recognition

Q:

What are the main benefits of Gaussian pyramids?

A:

1.) Solving problems in a coarse to fine manner accelerates most algorithms.

2.) Much more memory efficient as scale-spaces due to the downsampling step

Q:

Whats the kurtosis and skewness of a Gaussian?

A:

both 0

Q:

Are voxels usually cubic?

A:

No, the voxel dimensions usually differ in different directions. However, pixels are mostly squared.

Q:

Is 1 byte sufficent for the co-domain discretisation?

A:

Yes. As a human can only distinguish only around 40 greyscales, 256 are more than sufficient.

Q:

What are examples of vector valued images?

A:

1) Colour images RGB -> 3 dimensions

2) Multispectral images

Fun fact: humans can distinguish 2 million colours

Q:

Whats quantisation?

A:

Discretisation of the co-domain (wertebereich) --> value discretisation

Q:

What are main benefits of preprocessing?

A:

1.) Filtering: removal of noise and artifacts

2.) Simplification: makes it easier to interpret the content

3.) Scaling

4.) Rotating

..

Q:

What is regarded as concentration in a diffusion process in computer vision?

A:

grey values

Q:

Which classes of invariants do you know?

A:

geometric (e.g. translation, rotation, scaling)

### Erstelle und finde Lernmaterialien auf StudySmarter.

Greife kostenlos auf tausende geteilte Karteikarten, Zusammenfassungen, Altklausuren und mehr zu.

## Das sind die beliebtesten Computer Vision Kurse im gesamten StudySmarter Universum

##### computer

Silliman University

##### computer

University of Dohuk (Kurdistan Region)

##### Vision

Université Charles-de-Gaulle (Lille III)