Lecture 10

Question 1

Object classification/recognition

Answer 1

automatic labelling of objects based on visual characteristics

Question 2

Feature

Answer 2

those components of an image that contain useful information

Question 3

how to find good features:

Answer 3

1. have some degree of invariance (if things change in the image, the feature representation remain the same)
2. are discriminative (object from different categories have different features
3. are consistent (objects from the same category have similar feature)

Question 4

What features can be used for Feature detection

Answer 4

color histograms
corners
SIFTs

Question 5

SIFT descriptors

Answer 5

take a window 16x16 around the key point of interest compute the orientation of gradient at each pixel

Question 6

Application of SIFT

Answer 6

Matching key components of two images

Question 7

Scale-invariant feature transform (SIFT)

Answer 7

Detects point of interest and computing features descriptors for each point