Contextual guidance of eye movements and attention in real-world scenes: the role of global features in object search.

By Antonio Torralba, Aude Oliva, Monica S. Castelhano, J. Henderson. 2006

1,760 citations recorded by the source corpus.

Subfields: Cognitive Neuroscience, Computer Vision and Pattern Recognition, Neuroscience, Cognitive Science & Psychology

How researchers see this paper

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

In Cognitive Neuroscience

In Computer Vision and Pattern Recognition

In Neuroscience

In Cognitive Science & Psychology

Citations in the corpus

Recurrent Models of Visual Attention (2014, background): “Saliency detectors indeed capture some of the properties of human eye movements, but they typically do not to integrate information across fixations, their saliency computations are mostly hardwired, and they are based on low-level image properties only, usually ignoring other factors such as semant”
Comparing Automatic Evaluation Measures for Image Description (2014, background): “The underlying cause of this is an active area of research in the human vision literature and can be attributed to bottom-up effects, such as saliency (Itti et al., 1998), top-down contextual effects (Torralba et al., 2006), or rapidly-obtained scene properties (Oliva and Torralba, 2001).”
Recognizing indoor scenes (2009, methodology): “This notion of scene prototypes has been used in previous works [11, 17].”
Global-and-local attention networks for visual recognition (2018, background): “Decades of work on human attention have shown that our visual system uses at least two pathways to quickly guide attention to regions of interest [9]: A “global pathway” rapidly extracts a statistical summary of the whole scene in as little as a glance [10] while a complementary “local pathway” leve”
Learning Where to Attend with Deep Architectures for Image Tracking (2011, background): “These features can be computed from static images (Torralba et al., 2006), or from local regions of spacetime (Gaborski et al., 2004) for video.”
Towards Holistic Scene Understanding: Feedback Enabled Cascaded Classification Models (2010, neutral): “The feature inputs for the first-layer event classifier Ψ3 ∈ R(43) is a 43-dimensional feature vector, which includes the top 30 PCA projections of the 512dimensional GIST features [61], the 12-dimension global”
Best practices in eye tracking research. (2020, background): “Some of these are visual, and include the complexity and salience of the stimulus or its parts (Baddeley and Tatler, 2006; Itti and Koch, 2001; Nuthmann, 2017; Parkhurst et al., 2002; Torralba et al., 2006).”
Exploring set size effects in scenes: Identifying the objects of search (2008, background): “Although a negative result, this finding supports recent work showing a restriction of search to targetconsistent scene regions (Neider & Zelinsky, 2006b; see also Torralba et al., 2006, for a computational model, and Henderson et al., 1999, for early observations).”
The natural/man-made distinction is made before basic-level distinctions in scene gist processing (2010, background): “First, the current results use human subject data to test a central assumption of an influential computational model of scene gist recognition, the Spatial Envelope model (Oliva & Torralba, 2001, 2006; Oliva et al., 2003; Torralba, 2003; Torralba et al., 2006), which has been essentially untested.”
Learning to combine foveal glimpses with a third-order Boltzmann machine (2010, background): “A vast array of work has been dedicated to modelling the visual search behavior of humans [11, 12, 13, 14], typically through the computation of saliency maps [15, 16].”
Emergence of exploratory look-around behaviors through active observation completion (2019, background): “ate perception. For instance, humans rarely need to fully observe an object to understand its 3D shape (4–6), and one can often understand the primary contents of a room without literally scanning it (7). In short, given a set of past observations, some new views are more informative than others. Se”
Predicting the Driver's Focus of Attention: The DR(eye)VE Project (2017, background): “Usually, top-down computer vision models are built to integrate semantic contextual information in the attention prediction process [64].”
Task and context determine where you look. (2016, background): “Torralba et al. (2006) also considered an object detection task and proposed a spatial modulation of the saliency map by an additional map representing the likely positions of target objects in a scene.”
Learning to Look Around: Intelligently Exploring Unseen Environments for Unknown Tasks (2017, background): “For instance, humans rarely need to fully observe an object to understand its 3D shape [33,58,59], and one can often understand the primary contents of a room without literally scanning it [63].”