A collaborative benchmark for region of interest detection algorithms

Huang, Tsun‐Sheng; Cheng, Kai-Yin; Chuang, Yung-Yu

doi:10.1109/cvpr.2009.5206765

Cited by 21 publications

(14 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[Privitera and Stark 2000] compare region of interest algorithms based on how well they match eye fixations measured with an eye tracker. [Huang et al 2009] compare algorithms with respect to points collected in a game (Photoshoot) that asks people to select points on images that they expect will match a partner's selection, much like the ESP game proposed by [Von Ahn and Dabbish 2008]. This work is similar to ours in that it also asks people to match point selections.…”

Section: Introductionmentioning

confidence: 62%

“…Following the approach of [Chen et al 2009] and [Cole et al 2009], we recruited subjects for our study through Amazon's Mechanical Turk (AMT) [Amazon 2009], an on-line platform that matches people willing to work with paying tasks. Alternatively, we could have designed an on-line game to acquire input (as in [Huang et al 2009;Von Ahn and Dabbish 2008]), but that approach would have required attracting a player population, which is beyond the scope of this paper. Since tasks on the AMT are typically short in duration (a minute or two), inexpensive (around 10 cents), and accessible on-line (in a web page), it is well-suited for studies like ours that require lots of people to do simple, menial tasks (e.g., clicking points on a surface).…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Schelling points on 3D surface meshes

et al. 2012

View full text Add to dashboard Cite

This paper investigates "Schelling points" on 3D meshes, feature points selected by people in a pure coordination game due to their salience. To collect data for this investigation, we designed an online experiment that asked people to select points on 3D surfaces that they expect will be selected by other people. We then analyzed properties of the selected points, finding that: 1) Schelling point sets are usually highly symmetric, and 2) local curvature properties (e.g., Gauss curvature) are most helpful for identifying obvious Schelling points (tips of protrusions), but 3) global properties (e.g., segment centeredness, proximity to a symmetry axis, etc.) are required to explain more subtle features. Based on these observations, we use regression analysis to combine multiple properties into an analytical model that predicts where Schelling points are likely to be on new meshes. We find that this model benefits from a variety of surface properties, particularly when training data comes from examples in the same object class.

show abstract

Section: Introductionmentioning

confidence: 62%

Section: Methodsmentioning

confidence: 99%

Schelling points on 3D surface meshes

et al. 2012

View full text Add to dashboard Cite

show abstract

“…All data can be obtained trough http://www.cmlab.csie.ntu.edu.tw/roi/. The corresponding paper [15] compares various known computational saliency maps. Among these, the method presented in [26,27] provides the overall best performance, achieving a Spearman rank correlation with the ground truth saliency maps of 0.5784.…”

Section: Gradient Patchesmentioning

confidence: 99%

“…Its implementation is nonetheless relatively straightforward and its calculation needs less than double the amount of operations the Harris map requires. Experiments with our operator were run on the collaborative benchmark for region of interest detection [15], which provides a large collection of 993 images with corresponding region of interest ground truths derived from human annotations. All data can be obtained trough http://www.cmlab.csie.ntu.edu.tw/roi/.…”

Section: Gradient Patchesmentioning

confidence: 99%

“…One of the illustrative examples checks that with the same choice of features as in [28], the Harris map is indeed recovered. Another choice that considers patches of gradients, which is rather loosely inspired by the success of histogram of oriented gradient descriptors [10,4], is demonstrated to give state-of-the-art results on a collaborative benchmark for region of interest detection [15].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Information theoretic preattentive saliency: A closed-form solution

Loog

2011

2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops)

View full text Add to dashboard Cite

Employing an information theoretic operational definition of bottom-up attention from the field of computational visual perception, a very general expression for saliency is provided. As opposed to many of the current approaches to determining a saliency map, there is no need for an explicit, data-driven density estimation. Given the features, descriptors, or filter bank that one wants to use to describe the image content at every position, we provide a closed-form expression for the associated saliency at that location. This, indeed, makes explicit that what is considered salient depends on how, i.e., by means of which features, image information is described. We illustrate our result by determining a few specific saliency maps based on particular choices of features. One of them makes the link with the mapping underlying well-known Harris interest points, which is a result recently obtained in isolation. Another choice of features is, rather loosely, inspired by the success of histogram of oriented gradient descriptors and proves to provide stateof-the-art results on a collaborative benchmark for region of interest detection.

show abstract