THE ROLE OF LATEST FIXATIONS ON ONGOING VISUAL SEARCH
Download as PPT, PDF0 likes512 views
The document discusses a model evaluating the selection mechanism in visual search, distinguishing between top-down and bottom-up processes. It highlights the impact of recent fixations and their correlation with efficiency in task performance, specifically noting the importance of salience and the ability to inhibit distractions. Future research will focus on the relationship between these factors and the inhibition of return in visual processing.
![Task
“Visual Search depends
on sensory, perceptual
and cognitive processes”
[colby 2009]
Il TMT misuar la capcità
di esplorare ed eseguire
una sequenza in modo
corretto
TMTB è un mappa
viariante in cui i target
diventano distrattori
[Wolwer 2003]
07/01/13 www.evalab.unisi.it 3](https://image.slidesharecdn.com/raw2009-presentation-130107161012-phpapp01/85/THE-ROLE-OF-LATEST-FIXATIONS-ON-ONGOING-VISUAL-SEARCH-3-320.jpg)
THE ROLE OF LATEST FIXATIONS ON ONGOING VISUAL SEARCH
- 1. THE ROLE OF LATEST FIXATIONS ON ONGOING VISUAL SEARCH A MODEL TO EVALUATE THE SELECTION MECHANISM Giacomo Veneri, Pamela Federighi,Francesca Rosini, Elena Pretegiani, Antonio Federico, Alessandra Rufa Eye tracking & Vision Applications Lab (EVA Lab) Department of Neurological Neurosurgical and Behavioral Science University of Siena, Italy 07/01/13 www.evalab.unisi.it 1
- 2. Outline • Top-Down vs Bottom-Up on ongoing visual search • Efficiency on Visual search • (Bayesian) Model on visual search 07/01/13 www.evalab.unisi.it 2
- 3. Task “Visual Search depends on sensory, perceptual and cognitive processes” [colby 2009] Il TMT misuar la capcità di esplorare ed eseguire una sequenza in modo corretto TMTB è un mappa viariante in cui i target diventano distrattori [Wolwer 2003] 07/01/13 www.evalab.unisi.it 3
- 4. Mathematical Method Fixations 07/01/13 www.evalab.unisi.it 4
- 5. Direction vs Previous fixations In order to test an overall validity of the model, we calculated for each test the DEMINFIX(T) – DEMINFIX and the DEMAXFIX(T) – DEMAXFIX.Figures show that saccade direction take in consideration mainly fixations of last 1000ms (1s). 07/01/13 www.evalab.unisi.it 5
- 6. 07/01/13 www.evalab.unisi.it 6
- 7. Model 07/01/13 www.evalab.unisi.it 7
- 8. Human vs Machine 07/01/13 www.evalab.unisi.it 8
- 9. Conclusion • We found significant differences among tasks and a correlation between the efficiency (time elapsed) to explore the task and the ability to inhibit the BAF. • The command constraint (goal-driven) which is modulated by the image salience versus BAF. • The selection mechanism that drives this competition. Further works will be directed to evaluate the relation between the BAF and the inhibition of return. 07/01/13 www.evalab.unisi.it 9
- 10. Conclusion • We found significant differences among tasks and a correlation between the efficiency (time elapsed) to explore the task and the ability to inhibit the BAF. • The command constraint (goal-driven) which is modulated by the image salience versus BAF. • The selection mechanism that drives this competition. Further works will be directed to evaluate the relation between the BAF and the inhibition of return. 07/01/13 www.evalab.unisi.it 9