Lecture 11 Flashcards
-The ability to recognise and categorize objects is fundamental for survival and interaction with our environment
-It allows us to navigate our world, recognize dangers, find food, and plays a crucial role in social interactions
Why is recognising visual objects important?
-The variability of objects
-Their context
-Lighting conditions
What are the challenges in creating effective object recognition systems?
Template Theory
The proposition that the visual system recognizes objects by matching the neural representation of the image with an internal representation of the same ‘‘shape’’ in the brain
-But it would be very difficult for our brains to hold a template for each different ‘‘a’’ or cow
Exemplar Theory
Multiple stored examples
-Comparing new inputs to what you have already stored from the past
Prototype Theory
A single, abstracted prototype per category
-Matching to the most representative prototype
Generalized Context Model (Nosofsky)
You store many specific faces you’ve seen before. When seeing a new face, you compare it to stored examples and assign the category based on similarity to past faces
General Recognition Theory ( Ashby)
You rely on perceptual dimensions (e.g. face shape, jaw width, eye size) and make a decision based on statistical bounderies between the two categories
Recognition by Components (Biederman)
-Alphabet of shapes
-36 geons
-With these geons, you can draw any object
-Does not explain the variability between objects
Grandmother Cell Theory (Lettvin)
-Could a single neuron be responsible for recognizing your grandmother?
-This concept contributes to the ongoing debate between localized vs. distributed representation
A Jennifer Aniston Cell?
-Shown a picture of J.A. -> same cell keeps firing
-Didn’t work for look-alikes
-Didn’t work with others in pictures
Deep Neural Network (DNN)
-Multilayer neural networks capable of being trained to recognize objects
-Numerous instances of an object are shown to the network, with feedback provided
-Over time, the network learns to recognize new instances of the object that it has never been explicitly trained on
Retinal Ganglion Cells & LGN
Detect spots (localized contrast)
Primary Visual Cortex (V1)
Detect edges and bars (orientation selectivity)
Intermediate-Level Vision (V2, V3, V4, etc)
Grouping features into contours, textures, and surfaces
High-Level Vision (IT Cortex)
Recognizing complex shapes, objects, and categories
Intermediate (mid) Level Vision
A loosely defined stage of visual processing that occurs after low-level feature extraction (e.g. edges, contrast) and before high-level object recognition and scene understanding
Key Functions of Intermediate (mid) Level Vision
-Perception of edges and surfaces
-Determines which regions of an image should be grouped into objects
-Bridges low-level feature detection and high-level object recognition
-Primary visual cortex (V1) neurons have small receptive fields that detect local edges and contrast
-These neurons are orientation-selective, responding to edges at specific angles
How do we detect object edges?
The receptive fields of extrastriate cells respond to visual properties crucial for object perception
How do we know which edges belong together?
Boundery Ownership
For a given edge or contour, these neurons determine which side belongs to the object and which side belongs to the background - a fundamental process in figure-ground segregation
Computerized Edge Detectors
Not as effective as humans in detecting meaningful edges
-Miss edges that humans easily perceive because they rely purely on local contrast and intensity differences
Illusory Contour
A contour that is perceived even though no physical edge exists between one side and the other
Gestalt Theory
'’The whole is greater than the sum of its parts’’
-Opposes structuralism, which emphasizes breaking perception into basic elements
-Suggests that perception is holistic, meaning we naturally organize elements into meaningful wholes
Gestalt Grouping Principles
A set of rules that describe when and how elements in an image appear grouped together
Similarity
Similar objects (colour, shape, size) appear grouped
Proximity
Elements close to each other tend to be grouped
Good Continuation
Lines and edges are perceived as following the smoothest path
Closure
The mind fills in missing information to perceive complete shapes
Common Fate
Elements moving together are grouped
Figure Ground
The brain seperates objects from the background
Common Region
Elements located within a shared boundary or enclosed area are perceived as a group
Connectedness
Elements visually connected by lines tend to be grouped
Parallelism
Parallel contours are likely to belong to the same group
Symmetry
Symmetrical regions are more likely to be perceived as a group
Camouflage
Animals take advantage of Gestalt grouping principles to form groups in their environment
Ambiguity & Perceptual Committees
What are these Gestalt rules good for?
Perceptual Committees
-Committees must integrate conflicting inputs and reach a consencus
-Many different and sometimes competing principles influence perception
-Perception emerges as the result of the dominant interpretation agreed upon by these processes
Five Principles of Intermediate Vision
- Grouping what should be grouped together
- Separate what should be separated
- Use prior knowledge
- Avoid accidents
- Seek consencus and minimize ambiguity