Vision: A Computational Investigation into the Human Representation and Processing of Visual Information

By D. Marr. 1983

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

What can the brain teach us about building artificial intelligence? (2016, background): “Although Lake et al. seek to take the computational level ‘high ground’ (Marr, 1982), their representational commitments also constrain the inferential procedures they rely on.”
Self-Supervised Learning of Pretext-Invariant Representations (2019, background): “In fact, invariance is one of the core tenets of designing ‘good’ features [8, 45, 48].”
Explanation in Artificial Intelligence: Insights from the Social Sciences (2017, methodology): “Several other authors have proposed models similar to Aristotle’s, such as Dennett [35], who proposed that people take three stances towards objects: physical, design, and intention; and Marr [119], building on earlier work with Poggio [120], who define the computational, representational, and hardw”
Deep Unfolding: Model-Based Inspiration of Novel Deep Architectures (2014, methodology): “algorithms, or what David Marr called the “computational theory” level of analysis [1, 2].”
A theory of causal learning in children: causal maps and Bayes nets. (2004, background): “…vision systems) must also find procedures that allow them to compute the object representations from the retinal data in a reasonably efficient way Marr (1982) called this the algorithmic level of representation Vision scientists use the geometrical theory that relates depth to disparity to help…”
Specifying Object Attributes and Relations in Interactive Scene Generation (2019, background): “David Marr has defined vision as the process of discovering from images what is present in the world, and where it is [15].”
Improved Adversarial Systems for 3D Object Generation and Reconstruction (2017, background): “The pioneering work of Roberts [23] and Marr [21] exemplify early shape models composed of 3D geometric primitives with recognizable 2D properties.”
Recognition of natural scenes from global properties: Seeing the forest without representing the trees (2008, unknown): “Many influential models of high-level visual recognition are object-centered, treating objects and parts as the atoms of scene analysis (Biederman, 1987; Biederman, Blickle, Teitelbaum, Klatsky, & Mezzanotte, 1988; Bulthoff, Edelman, & Tarr, 1995; Fergus, Perona, & Zisserman, 2003; Marr, 1982; Pyl”
Using goal-driven deep learning models to understand sensory cortex (2016, result): “In this sense, the current work is more similar in spirit to Marr’s levels of analysis 71 , investigating how a system’s computational-level goals influence its algorithmic and implementation level mechanisms.”
Computing and Rendering Point Set Surfaces (2003, background): “Note that our visual system cannot sense smoothness beyond second order [38].”
Neural and Computational Mechanisms of Action Processing: Interaction between Visual and Motor Representations. (2015, background): “This conception of different levels of theory in neuroscience matches the classical distinction of levels of analysis that has been proposed by Marr (1982) for computational vision.”