Components of Bottom-Up Gaze Allocation in Natural Scenes

Components of Bottom-Up Gaze Allocation in Natural Scenes. Peters, R. J., Iyer, A., Koch, C., & Itti, L. In Proc. Vision Science Society Annual Meeting (VSS05), May, 2005.
abstract bibtex

A model of bottom-up visual attention ("baseline salience model", based on local detectors with coarse global surround inhibition) has been shown (Parkhurst et al., 2002) to account in part for the spatial locations fixated by people while free-viewing complex natural and artificial scenes. Here, we tested the additional roles in bottom-up gaze allocation played by several visual cortical mechanisms. In each case, we added a component to the salience model: non-linear interactions among orientation-tuned units both at short spatial ranges (for clutter reduction) and long ranges (for contour facilitation), and a detailed model of eccentricity-dependent changes in visual processing. Subjects free-viewed naturalistic and artificial images while their eye movements were recorded, and we used a metric called the Normalized Scanpath Salience (NSS) to compare the resulting fixation locations with the different models' predicted salience maps. NSS values indicate, on average, how many standard deviations above or below the mean salience was the model-predicted salience at human-fixated locations. Thus the minimum NSS value (when the model and human behavior are unrelated) is 0; the theoretical maximum NSS value is given by the ability of one observer's fixations to be predicted by the remaining observers' fixations, which in practice fell in the range 1.1--1.3 for different image categories. The baseline salience model predicted fixations at 39--57 percent of the maximum NSS level. Adding short-range orientation interactions increased this range to 50--65 percent, contour facilitation further increased it to 53--74 percent, and eccentricity-dependent processing increased it to 84--95 percent. Thus the proposed cortical interactions indeed appear to play a significant role in the spatiotemporal deployment of attention in natural scenes. This suggests that bottom-up attentional guidance does not depend solely on local visual features, but must also include the effects of non-local interactions.

@inproceedings{ Peters_etal05vss,
  author = {R. J. Peters and A. Iyer and C. Koch and L. Itti},
  title = {Components of Bottom-Up Gaze Allocation in Natural Scenes},
  abstract = {A model of bottom-up visual attention ("baseline salience
model", based on local detectors with coarse global surround
inhibition) has been shown (Parkhurst et al., 2002) to account in part
for the spatial locations fixated by people while free-viewing complex
natural and artificial scenes. Here, we tested the additional roles in
bottom-up gaze allocation played by several visual cortical
mechanisms. In each case, we added a component to the salience model:
non-linear interactions among orientation-tuned units both at short
spatial ranges (for clutter reduction) and long ranges (for contour
facilitation), and a detailed model of eccentricity-dependent changes
in visual processing. Subjects free-viewed naturalistic and artificial
images while their eye movements were recorded, and we used a metric
called the Normalized Scanpath Salience (NSS) to compare the resulting
fixation locations with the different models' predicted salience
maps. NSS values indicate, on average, how many standard deviations
above or below the mean salience was the model-predicted salience at
human-fixated locations. Thus the minimum NSS value (when the model
and human behavior are unrelated) is 0; the theoretical maximum NSS
value is given by the ability of one observer's fixations to be
predicted by the remaining observers' fixations, which in practice
fell in the range 1.1--1.3 for different image categories. The
baseline salience model predicted fixations at 39--57 percent of the
maximum NSS level. Adding short-range orientation interactions
increased this range to 50--65 percent, contour facilitation further
increased it to 53--74 percent, and eccentricity-dependent processing
increased it to 84--95 percent. Thus the proposed cortical
interactions indeed appear to play a significant role in the
spatiotemporal deployment of attention in natural scenes. This
suggests that bottom-up attentional guidance does not depend solely on
local visual features, but must also include the effects of non-local
interactions.},
  booktitle = {Proc. Vision Science Society Annual Meeting (VSS05)},
  year = {2005},
  month = {May},
  type = {mod;bu;eye},
  review = {abs/conf}
}

Downloads: 0

{"_id":{"_str":"5298a1a09eb585cc26000844"},"__v":0,"authorIDs":[],"author_short":["Peters, R.<nbsp>J.","Iyer, A.","Koch, C.","Itti, L."],"bibbaseid":"peters-iyer-koch-itti-componentsofbottomupgazeallocationinnaturalscenes-2005","bibdata":{"html":"<div class=\"bibbase_paper\"> \n\n\n<span class=\"bibbase_paper_titleauthoryear\">\n\t<span class=\"bibbase_paper_title\"><a name=\"Peters_etal05vss\"> </a>Components of Bottom-Up Gaze Allocation in Natural Scenes.</span>\n\t<span class=\"bibbase_paper_author\">\nPeters, R. J.; Iyer, A.; Koch, C.; and Itti, L.</span>\n\t\n</span>\n\n\n\nIn\n<i>Proc. Vision Science Society Annual Meeting (VSS05)</i>, May 2005.\n\n\n\n\n\n<br class=\"bibbase_paper_content\"/>\n\n<span class=\"bibbase_paper_content\">\n \n \n \n <a href=\"javascript:showBib('Peters_etal05vss')\"\n class=\"bibbase link\">\n \n\t\n\t\n\t\n BibTeX\n <i class=\"fa fa-caret-down\"></i></a>\n \n \n  \n <a class=\"bibbase_abstract_link bibbase link\"\n href=\"javascript:showAbstract('Peters_etal05vss')\">\n Abstract\n <i class=\"fa fa-caret-down\"></i></a>\n \n \n \n\n \n \n \n</span>\n\n<div class=\"well well-small bibbase\" id=\"bib_Peters_etal05vss\"\n style=\"display:none\">\n <pre>@inproceedings{ Peters_etal05vss,\n author = {R. J. Peters and A. Iyer and C. Koch and L. Itti},\n title = {Components of Bottom-Up Gaze Allocation in Natural Scenes},\n abstract = {A model of bottom-up visual attention (\"baseline salience\nmodel\", based on local detectors with coarse global surround\ninhibition) has been shown (Parkhurst et al., 2002) to account in part\nfor the spatial locations fixated by people while free-viewing complex\nnatural and artificial scenes. Here, we tested the additional roles in\nbottom-up gaze allocation played by several visual cortical\nmechanisms. In each case, we added a component to the salience model:\nnon-linear interactions among orientation-tuned units both at short\nspatial ranges (for clutter reduction) and long ranges (for contour\nfacilitation), and a detailed model of eccentricity-dependent changes\nin visual processing. Subjects free-viewed naturalistic and artificial\nimages while their eye movements were recorded, and we used a metric\ncalled the Normalized Scanpath Salience (NSS) to compare the resulting\nfixation locations with the different models' predicted salience\nmaps. NSS values indicate, on average, how many standard deviations\nabove or below the mean salience was the model-predicted salience at\nhuman-fixated locations. Thus the minimum NSS value (when the model\nand human behavior are unrelated) is 0; the theoretical maximum NSS\nvalue is given by the ability of one observer's fixations to be\npredicted by the remaining observers' fixations, which in practice\nfell in the range 1.1--1.3 for different image categories. The\nbaseline salience model predicted fixations at 39--57 percent of the\nmaximum NSS level. Adding short-range orientation interactions\nincreased this range to 50--65 percent, contour facilitation further\nincreased it to 53--74 percent, and eccentricity-dependent processing\nincreased it to 84--95 percent. Thus the proposed cortical\ninteractions indeed appear to play a significant role in the\nspatiotemporal deployment of attention in natural scenes. This\nsuggests that bottom-up attentional guidance does not depend solely on\nlocal visual features, but must also include the effects of non-local\ninteractions.},\n booktitle = {Proc. Vision Science Society Annual Meeting (VSS05)},\n year = {2005},\n month = {May},\n type = {mod;bu;eye},\n review = {abs/conf}\n}</pre>\n</div>\n\n\n<div class=\"well well-small bibbase\" id=\"abstract_Peters_etal05vss\"\n style=\"display:none\">\n A model of bottom-up visual attention (\"baseline salience model\", based on local detectors with coarse global surround inhibition) has been shown (Parkhurst et al., 2002) to account in part for the spatial locations fixated by people while free-viewing complex natural and artificial scenes. Here, we tested the additional roles in bottom-up gaze allocation played by several visual cortical mechanisms. In each case, we added a component to the salience model: non-linear interactions among orientation-tuned units both at short spatial ranges (for clutter reduction) and long ranges (for contour facilitation), and a detailed model of eccentricity-dependent changes in visual processing. Subjects free-viewed naturalistic and artificial images while their eye movements were recorded, and we used a metric called the Normalized Scanpath Salience (NSS) to compare the resulting fixation locations with the different models' predicted salience maps. NSS values indicate, on average, how many standard deviations above or below the mean salience was the model-predicted salience at human-fixated locations. Thus the minimum NSS value (when the model and human behavior are unrelated) is 0; the theoretical maximum NSS value is given by the ability of one observer's fixations to be predicted by the remaining observers' fixations, which in practice fell in the range 1.1--1.3 for different image categories. The baseline salience model predicted fixations at 39--57 percent of the maximum NSS level. Adding short-range orientation interactions increased this range to 50--65 percent, contour facilitation further increased it to 53--74 percent, and eccentricity-dependent processing increased it to 84--95 percent. Thus the proposed cortical interactions indeed appear to play a significant role in the spatiotemporal deployment of attention in natural scenes. This suggests that bottom-up attentional guidance does not depend solely on local visual features, but must also include the effects of non-local interactions.\n</div>\n\n\n</div>\n","downloads":0,"bibbaseid":"peters-iyer-koch-itti-componentsofbottomupgazeallocationinnaturalscenes-2005","role":"author","year":"2005","type":"mod;bu;eye","title":"Components of Bottom-Up Gaze Allocation in Natural Scenes","review":"abs/conf","month":"May","key":"Peters_etal05vss","id":"Peters_etal05vss","booktitle":"Proc. Vision Science Society Annual Meeting (VSS05)","bibtype":"inproceedings","bibtex":"@inproceedings{ Peters_etal05vss,\n author = {R. J. Peters and A. Iyer and C. Koch and L. Itti},\n title = {Components of Bottom-Up Gaze Allocation in Natural Scenes},\n abstract = {A model of bottom-up visual attention (\"baseline salience\nmodel\", based on local detectors with coarse global surround\ninhibition) has been shown (Parkhurst et al., 2002) to account in part\nfor the spatial locations fixated by people while free-viewing complex\nnatural and artificial scenes. Here, we tested the additional roles in\nbottom-up gaze allocation played by several visual cortical\nmechanisms. In each case, we added a component to the salience model:\nnon-linear interactions among orientation-tuned units both at short\nspatial ranges (for clutter reduction) and long ranges (for contour\nfacilitation), and a detailed model of eccentricity-dependent changes\nin visual processing. Subjects free-viewed naturalistic and artificial\nimages while their eye movements were recorded, and we used a metric\ncalled the Normalized Scanpath Salience (NSS) to compare the resulting\nfixation locations with the different models' predicted salience\nmaps. NSS values indicate, on average, how many standard deviations\nabove or below the mean salience was the model-predicted salience at\nhuman-fixated locations. Thus the minimum NSS value (when the model\nand human behavior are unrelated) is 0; the theoretical maximum NSS\nvalue is given by the ability of one observer's fixations to be\npredicted by the remaining observers' fixations, which in practice\nfell in the range 1.1--1.3 for different image categories. The\nbaseline salience model predicted fixations at 39--57 percent of the\nmaximum NSS level. Adding short-range orientation interactions\nincreased this range to 50--65 percent, contour facilitation further\nincreased it to 53--74 percent, and eccentricity-dependent processing\nincreased it to 84--95 percent. Thus the proposed cortical\ninteractions indeed appear to play a significant role in the\nspatiotemporal deployment of attention in natural scenes. This\nsuggests that bottom-up attentional guidance does not depend solely on\nlocal visual features, but must also include the effects of non-local\ninteractions.},\n booktitle = {Proc. Vision Science Society Annual Meeting (VSS05)},\n year = {2005},\n month = {May},\n type = {mod;bu;eye},\n review = {abs/conf}\n}","author_short":["Peters, R.<nbsp>J.","Iyer, A.","Koch, C.","Itti, L."],"author":["Peters, R. J.","Iyer, A.","Koch, C.","Itti, L."],"abstract":"A model of bottom-up visual attention (\"baseline salience model\", based on local detectors with coarse global surround inhibition) has been shown (Parkhurst et al., 2002) to account in part for the spatial locations fixated by people while free-viewing complex natural and artificial scenes. Here, we tested the additional roles in bottom-up gaze allocation played by several visual cortical mechanisms. In each case, we added a component to the salience model: non-linear interactions among orientation-tuned units both at short spatial ranges (for clutter reduction) and long ranges (for contour facilitation), and a detailed model of eccentricity-dependent changes in visual processing. Subjects free-viewed naturalistic and artificial images while their eye movements were recorded, and we used a metric called the Normalized Scanpath Salience (NSS) to compare the resulting fixation locations with the different models' predicted salience maps. NSS values indicate, on average, how many standard deviations above or below the mean salience was the model-predicted salience at human-fixated locations. Thus the minimum NSS value (when the model and human behavior are unrelated) is 0; the theoretical maximum NSS value is given by the ability of one observer's fixations to be predicted by the remaining observers' fixations, which in practice fell in the range 1.1--1.3 for different image categories. The baseline salience model predicted fixations at 39--57 percent of the maximum NSS level. Adding short-range orientation interactions increased this range to 50--65 percent, contour facilitation further increased it to 53--74 percent, and eccentricity-dependent processing increased it to 84--95 percent. Thus the proposed cortical interactions indeed appear to play a significant role in the spatiotemporal deployment of attention in natural scenes. This suggests that bottom-up attentional guidance does not depend solely on local visual features, but must also include the effects of non-local interactions."},"bibtype":"inproceedings","biburl":"http://ilab.usc.edu/publications/src/ilab.bib","downloads":0,"search_terms":["components","bottom","gaze","allocation","natural","scenes","peters","iyer","koch","itti"],"title":"Components of Bottom-Up Gaze Allocation in Natural Scenes","year":2005,"dataSources":["wedBDxEpNXNCLZ2sZ"]}