Object Recognition Flashcards
Object Recognition
process through which an object is recognised.
Form Perception
Process through which basic shape and size of an object is established.
Laws of Pragnanz
Proximity, good continuation, similarity, simplicity and closure.
Word Superiority Effect
Words are recognised faster than strings of random letters.
Feature Nets
Letter recognition depends on detectors for features (parts of letters) that are located in complex bundles of neurons.
Support: provides explanation for recognition errors
Limitations: sacrificing accuracy for efficiency.
Recognition by Components
Shapes are detected, the components of the objects are determined as geons, the geons are combined to form complex objects.
Support: it takes longer to recognise incomplete or complex shapes.
Limitations: deals with block shapes rather than real life objects.
Recognition by Multiple Views
There is a stored number of views of an object so the object may need to be rotated in order to perceive it.
Support: recognition occurs faster if the object is viewed from a certain angle and there are some neurons in the brain that fire more rapidly when views align
Limitations: slower form of recognition
Facial Recognition
Viewpoint dependent, involves identifying expressions, integrating knowledge about the person and recognising facial units.
Support: parts of the brain are particularly sensitive to faces and faces look distorted if the orientation is wrong.
Limitations: the same brain area for faces is also used when distinguishing between similar objects (eg cars or birds)
Bottom-up Processing
Data driven. Based on information coming in and building up to a complete object. Feature nets are bottom-up
Top-down Processing
Concept driven. Based on broad knowledge and expectations which triggers detectors.
