Object Recognition Flashcards Preview

Perception > Object Recognition > Flashcards

Flashcards in Object Recognition Deck (30)
Loading flashcards...
1

2 pathways of the brain

ventral and dorsal

2

what is object agnosia

intact primary visual functions, verbal descriptions and logical reasoning but immediate object recognition is lost

3

which stream is damaged in object agnosia

ventral

4

What kind of approach is Marr's model of object recognition/vision

computational

5

What does Marr's model of vision suggest

1. vision is an information processing task
2. we have to understand the nature of the task
3. we have to understand how it can be accomplished
4. we need to propose algorithms and mechanisms that can accomplish the visual task
5. explanations of visual experience and visual physiology should come from an understanding of the implementation of those algorithms

6

Different models of object recognition used

template-matching model
feature-detector models
structural-description models
view-dependent models

7

template-matching model

match the exact pattern of light on the retina with previously remembered patterns.
Neuron makes connections with pixels and when the specific pixels are detected the neuron fires

8

problems with template-matching model

1. very constrained - can't cope with diversity of natural object variation
2. doesn't explain how we detect variances in images
3. there would be too many template detectors

9

what model solves the problems of the template-matching model

feature-detector model: looks at specific features in the image rather than pixels

10

Who put forward the idea of 'feature-detectors'?

Hubel and Wiesel

11

what do feature-detector models suggest?

we can detect specific combinations of features rather than specific patterns
the features don't have to exactly match a pre-existing template, it just needs to contain the same sub-features that

12

what are demons in the feature-detector model

subroutines

13

who out forward the structural description model

Marr & Nishihara (1978)

14

What did Marr & Nishihara (1978) suggest the goal of vision is in the structural description model

to describe the object unambiguously in its core geometric components

15

Marr & Nishihara's critera for good representation of high level vision

1. Accessibility
2. scope
3. uniqueness
4. stability
5. sensitivity

16

what did Marr & Nishihara (1978) suggest the primitives (basic units of information in representation) are

objects are described in terms of axes and the volumes around them
the description is hierarchical
this means the object can be described at many scales

17

according to Marr & Nishihara (1978) how is information organised into an abject description?

recognition:
by finding the closest model in the model store and specifying its parameters

18

What did Biederman (1987) suggest for structural description models

similar to Marr & Nishihara (1978) but proposed a specific "alphabet" of primitive volumes callled Geons which object are decomposed

19

what properties are geons similar to

any 2D projection

20

how many geons are there estimated to be

less that 36

21

evidence for structural models

intersections of geons are particularly informative in object recognition e.g. it is hard to tell what an object is when straight lines or corners are removed

22

what are IT cells

higher visual neurons in IT area

23

Pros of structural description models

explain invariance well
recognition is description not matching
evidence that structural information matters to both humans and neurons

24

cones of structural description models

extracting model parameters can be hard in real mages
structural description is difficult for some objects e.g. campfires
driven by theoretical desirability rather than behavioural evidence

25

what does the the view-dependent model suggest

that the arbitrary image has a brute association with the object

26

what are the primitives in the view-dependent model?

sub-regions of the image
consisting of lines, curves, textures, colour, shading etc

27

what kind of model is the view-based model

simple feet forward activation

28

evidence for view-baed models

human object recognition is not perfectly viewpoint invariant
Monkey IT neurons selective for particular viewpoints
simulations show good invariance and similar errors in human observers

29

pros of view-dependent model

straightforward
newer models based directly on what we know of physiology
abstract feature units are recombines
good evidence

30

cons of view-based models

humans often show good generalisation across viewpoints even for novel objects
more memory intensive than other models e.g. geon models
no understanding of underlying relationships