狠狠撸

狠狠撸Share a Scribd company logo

Visual Perception Analysis with GazeHits

?
?
Jorge Rodríguez Araújo
Jorge Rodríguez Araújo

GazeHits is a design assistant tool to know how your design is perceived in terms of visual attention. It bases itself on state-of-the-art visual attention computer models to identify the elements and features that captures attention. GazeHits makes easy to analyze alternative designs and decide when a work is ready to present it to your customer.

1 of 24
Downloaded 15 times
Visual Perception Analysis with GazeHits
2
Visual perception analysis 1. What is GazeHits 2. GazeHits for conversion optimization 3. Science behind 4. Technology under GazeHits
● Qualitative ? Expert opinion ● Eye tracking ? Is costly and time consuming ? Requires hiring tens of people to acquire reliable data ? Does not trace the drivers of attention CC BY 2.0 SMI - Neil Rickards Conventional visual perception analysis
GazeHits, beyond eye tracking ● Immediate and Mobile ● Predicts first 3 seconds of eye tracking with high accuracy ● Predicts visual clutter perception with high correlation (0 to 100) ● Provides insight in why something is eye-catching or cluttered ● Provides quantitative indicators to be integrated with analytics
Importance of visual perception in marketing ● “If consumers’ attention is not caught in the first place, all findings [on message appreciation and acceptance] in ‘forced exposure’ research are valueless.” Wedel and Pieters. Review of marketing research. (2008): 123-147. ● “The density of visual detail in terms of color, luminance, and edges hurts brand attention and attitude toward the ad” Pieters, Wedel, and Batra. "The stopping power of advertising: Measures and effects of visual complexity." Journal of Marketing (2010): 48-60.
● Just upload and image ? from drive ? from camera shot ● Functionality ? Heatmap ? Opacity map ● Quantitative strength ? Region selection ? Visual clutter www.gazehits.com FREE on Beta How it works?
Visual perception analysis 1. What is GazeHits 2. GazeHits for conversion optimization 3. Science behind 4. Technology under GazeHits
Pre-testing in web design
Pre-testing web designs in e-commerce
Pre-testing ads in online newspapers
In context visual attention assessment
In context windows assessment
Visual perception analysis 1. What is GazeHits 2. GazeHits for conversion optimization 3. Science behind 4. Technology under GazeHits
● Acquires 10-100 Mb/s through retinal photoreceptors ● Our eyes move all the time: Saccades and Fixations ? Otherwise photoreceptor adaptation would make us blind in seconds ● Just one degree of visual field is perceived with high acuity at a time ? That part under the spot of the Fovea ● Certain stimuli receive enhanced neural processing ● Fixations ? Keep the fovea around a fixed point during 0.5s ? 2-3 fixations per second 4 facts about human visual system...
What is visual attention? ● Roughly, the set of operations that select visual information ● Key role in ? Building visual concepts to represent reality ? Object perception ? Focusing cognitive effort on important stimuli
● Biases of the visual system: e.g. looking at the center ? Unrelated to the specific image ● Top-down Relevance related to task, intention, and knowledge of observer ? Dependent on observer motivation and experience, generally divergent ● Bottom-up Saliency arising from the physical features of the scene ? Universal and common to any observer with healthy vision What are the drivers of attention?
Pop-out fenomena
Visual perception analysis 1. What is GazeHits 2. GazeHits for conversion optimization 3. Science behind 4. Technology under GazeHits
Models of attention ● Bioinspired ? Decompose and analyse images following observed operations in the human visual system ? Implement hypothesis on neural operations ● Traceability is possible ● Robust performance across different types of images ● Unaware of semantics or task ● Machine learning ? Learn to reproduce fixations registered in eye tracking experiments ● Capable to learn semantics and task-related behavior ● Reproduce biases in training data set ● Black box. No traceability
AWS, the model under GazeHits ● Based on the image representation by the visual cortex ? Hubel and Wisel, Nobel Prize of Medicine 1981 ● Based on statistical operations shown by the human visual system ? Decorrelation and normalization contrast
Technology under GazeHits ● AWS as the bioinspired component ? Accurate and robust to biases and changes in image types ? Ranked first with sAUC (no biases) in MIT CAT2000, the largest benchmark available ● Machine learning used for task-related prediction ? Can be trained with conventional analytics data
Competitors ● Feng-Gui (https://www.feng-gui.com/) ? Based on some kind of algorithms ● Attention Wizard (https://www.attentionwizard.com/) ? Based on algorithms to provide web heatmaps ● Eyequant (http://www.eyequant.com/) ? Based on machine learning (only for webs) ● 3M Visual Attention Service (http://solutions.3m.com/wps/portal/3M/en _US/VAS_NA/Home/)
www.gazehits.com info@gazehits.com Let us hear from your specific needs

More Related Content

Visual Perception Analysis with GazeHits

  • 2. 2
  • 3. Visual perception analysis 1. What is GazeHits 2. GazeHits for conversion optimization 3. Science behind 4. Technology under GazeHits
  • 4. ● Qualitative ? Expert opinion ● Eye tracking ? Is costly and time consuming ? Requires hiring tens of people to acquire reliable data ? Does not trace the drivers of attention CC BY 2.0 SMI - Neil Rickards Conventional visual perception analysis
  • 5. GazeHits, beyond eye tracking ● Immediate and Mobile ● Predicts first 3 seconds of eye tracking with high accuracy ● Predicts visual clutter perception with high correlation (0 to 100) ● Provides insight in why something is eye-catching or cluttered ● Provides quantitative indicators to be integrated with analytics
  • 6. Importance of visual perception in marketing ● “If consumers’ attention is not caught in the first place, all findings [on message appreciation and acceptance] in ‘forced exposure’ research are valueless.” Wedel and Pieters. Review of marketing research. (2008): 123-147. ● “The density of visual detail in terms of color, luminance, and edges hurts brand attention and attitude toward the ad” Pieters, Wedel, and Batra. "The stopping power of advertising: Measures and effects of visual complexity." Journal of Marketing (2010): 48-60.
  • 7. ● Just upload and image ? from drive ? from camera shot ● Functionality ? Heatmap ? Opacity map ● Quantitative strength ? Region selection ? Visual clutter www.gazehits.com FREE on Beta How it works?
  • 8. Visual perception analysis 1. What is GazeHits 2. GazeHits for conversion optimization 3. Science behind 4. Technology under GazeHits
  • 14. Visual perception analysis 1. What is GazeHits 2. GazeHits for conversion optimization 3. Science behind 4. Technology under GazeHits
  • 15. ● Acquires 10-100 Mb/s through retinal photoreceptors ● Our eyes move all the time: Saccades and Fixations ? Otherwise photoreceptor adaptation would make us blind in seconds ● Just one degree of visual field is perceived with high acuity at a time ? That part under the spot of the Fovea ● Certain stimuli receive enhanced neural processing ● Fixations ? Keep the fovea around a fixed point during 0.5s ? 2-3 fixations per second 4 facts about human visual system...
  • 16. What is visual attention? ● Roughly, the set of operations that select visual information ● Key role in ? Building visual concepts to represent reality ? Object perception ? Focusing cognitive effort on important stimuli
  • 17. ● Biases of the visual system: e.g. looking at the center ? Unrelated to the specific image ● Top-down Relevance related to task, intention, and knowledge of observer ? Dependent on observer motivation and experience, generally divergent ● Bottom-up Saliency arising from the physical features of the scene ? Universal and common to any observer with healthy vision What are the drivers of attention?
  • 19. Visual perception analysis 1. What is GazeHits 2. GazeHits for conversion optimization 3. Science behind 4. Technology under GazeHits
  • 20. Models of attention ● Bioinspired ? Decompose and analyse images following observed operations in the human visual system ? Implement hypothesis on neural operations ● Traceability is possible ● Robust performance across different types of images ● Unaware of semantics or task ● Machine learning ? Learn to reproduce fixations registered in eye tracking experiments ● Capable to learn semantics and task-related behavior ● Reproduce biases in training data set ● Black box. No traceability
  • 21. AWS, the model under GazeHits ● Based on the image representation by the visual cortex ? Hubel and Wisel, Nobel Prize of Medicine 1981 ● Based on statistical operations shown by the human visual system ? Decorrelation and normalization contrast
  • 22. Technology under GazeHits ● AWS as the bioinspired component ? Accurate and robust to biases and changes in image types ? Ranked first with sAUC (no biases) in MIT CAT2000, the largest benchmark available ● Machine learning used for task-related prediction ? Can be trained with conventional analytics data
  • 23. Competitors ● Feng-Gui (https://www.feng-gui.com/) ? Based on some kind of algorithms ● Attention Wizard (https://www.attentionwizard.com/) ? Based on algorithms to provide web heatmaps ● Eyequant (http://www.eyequant.com/) ? Based on machine learning (only for webs) ● 3M Visual Attention Service (http://solutions.3m.com/wps/portal/3M/en _US/VAS_NA/Home/)