Object and face recognition

• Gibson's (1979) “ecological” approach focuses on how individuals perceive and interact with their environment.
• Gibson's approach emphasizes the importance of perception in understanding behaviour.
• The optic array contains unambiguous information about the layout of objects in space, in the form of texture gradients, optic flow patterns etc.
• Perception involves “picking up” the rich information provided by the optic array directly via “resonance” - our prior experiences, beliefs, and emotions influence the way we perceive and interpret sensory information.
• A pilot approaching a landing strip experiences a point in their visual field appearing to remain stationary, while all the other points seem to be flowing away, providing direct information about the point the pilot is heading to.
• Gibson introduced the notion of direct perception of the affordances of an object, for example, a chair we perceive its “affordance” as something to sit on.
• Gibson’s approach highlights the richness of information present in the visual signal.
• Gibson’s approach acknowledges the environmental and ecological grounding of perception, stating that perception is not a passive process but an active engagement with the environment.
• Gibson’s approach transparently addresses the role of perception in guiding our actions and how we adapt to our surroundings.
• Gibson’s approach provides a holistic understanding of how individuals interact with their environment.
• Gibson’s approach oversimplifies complex cognitive processes by focusing primarily on perception and action.
• Gibson’s approach can be challenging to quantify and measure objectively as it relies heavily on subjective experiences and individual interpretations.
• Gibson’s approach avoids the problem that information processing accounts sought to address by not specifying what constitutes ‘resonance’ and ‘invariance detection’.
• Gibson’s approach denies the role of stored knowledge (memories) in perceptions, which is highly questionable.
• Gregory’s (constructivist) explanation of the Mueller-Lyer illusion is that the influence of cultural and environmental factors on our perception can explain the Mueller-Lyer illusion.
• Gregory argues that the presence of arrow like fins at the ends of the lines in the Mueller-Lyer illusion creates conflicting depth cues, leading us to perceive one line as being longer although they’re the same length.
• Gregory suggests that if the figures/corner in the Mueller-Lyer illusion look the same size but are at different distances from the observer, then the one that looks like an inside corner must be longer.
• Constructivists theories of object recognition suggest that perception is not directly given by the stimulus input but occurs as the end-product of the interactive influences such as expectations, emotional factors etc.
• Constructivists theories of object recognition argue that perception is not a passive reflection of the external world but an active and constructive process of interpretation and inference.
• Marr (1982) suggested a series of representations leading up to the recognition of objects in a visual scene.
• According to Gibson's ecological approach, what is the role of the optic array in perception? It contains unambiguous information about the arrangement of objects in space
• In Gibson's ecological approach, what does "resonance" refer to? The influence of prior experiences, beliefs, and emotions on perception.
• According to Gibson, what does the optic array contain?Unambiguous information
• What does Gibson's ecological approach emphasize in understanding behaviour? Importance of perception and interaction with the environment
• What is the concept of "affordances" in Gibson's ecological approach? Potential uses of an object based on its properties and our capabilities
• Marr's levels of perception include Grey level description, Raw primal sketch, Full primal sketch, 2½ -D sketch, and 3-D model.
• Grey level description simplifies images by removing colour information and represents different shades of grey based on the intensity of light in each pixel.
• Raw primal sketch refers to the initial stage of visual processing where basic features, such as edges, orientations, and contrasts, are extracted from the visual input.
• Full primal sketch involves extracting more complex visual features, such as contours, surfaces, and depth, from the raw primal sketch that go together to form larger structures in the image.
• 2½ -D sketch refers to a visual representation that captures the depth and structure of a 3D scene using 2D information, adds information about depth and orientation of visible surfaces, by using texture, motion, etc.
• Marr's model suggests that the brain processes visual information in multiple stages, starting with a primal sketch, then a 2.5D sketch, and finally a full 3D model.
• Marr and Nishihara (1978) suggested using cylinders as basic elements in these 3D model descriptions, where each cylinder’s long axis is appropriately aligned to capture the structure and orientation of objects in the scene.
• Biederman’s (1987) recognition by components model is a development of Marr and Nishihara’s (1978) theory.
• Biederman proposed a more general set of elements which he called geons (about 36), and images are broken down into geon components.
• At sharply concave parts is where breaks tend to be made in the description of the relationship between cylinders.
• Points or segments in a curve strongly imply a curve on the object in Biederman’s theory.
• Points in a straight line in a sketch strongly suggest that those points also lie in a straight line, but sometimes this assumption is wrong.
• Symmetrical parts of the sketch are highly likely to be symmetrical parts of the 3-D model in Biederman’s theory.
• Parallel parts of the sketch strongly imply that the corresponding parts of the object are parallel in Biederman’s theory.
• Once the objects in a scene are given these view-invariant descriptions, they can be identified by referring to objects stored in memory with the same or similar descriptions.