Ideas on paper content #18

New issue

Closed

opened 2018-09-04 10:18:19 +00:00 by mih · 2 comments

mih commented

2018-09-04 10:18:19 +00:00

(Migrated from github.com)

start with problem: low-quality (but precious) data from natural viewing (movies)
outline our algorithm idea
compare to ultimate comparison (see #16)
show performance on MRI eyegaze data
compare to HQ lab eyegaze data
link code repo with DOI from zenodo
link data repo
put in brainhack collection

- start with problem: low-quality (but precious) data from natural viewing (movies) - outline our algorithm idea - compare to ultimate comparison (see #16) - show performance on MRI eyegaze data - compare to HQ lab eyegaze data - link code repo with DOI from zenodo - link data repo - put in brainhack collection

adswa commented

2018-09-10 14:13:57 +00:00

(Migrated from github.com)

mean fixation durations in seconds for every subject (averaged across runs):

median fixation durations in seconds for every subject (across all runs):

This paper examined eye movements during natural viewing (incl "Hollywood trailers" (different Startwars movies)): Dorr, M., Martinetz, T., Gegenfurtner, K. R. & Barth, E. (2010). Variability of eye movements when viewing dynamic natural scenes. Journal of Vision, 10(10), 1 – 17. (https://jov.arvojournals.org/article.aspx?articleid=2121333) They find: mean: 354ms, median: 253ms.

Our overall mean and median (across all subj and runs): overall mean and median 0.326750136304 sec, 0.256 sec. --> really close.

(previously, these were the mean and median durations from the old output (the metric is ms):

which overall amounted to a mean of 444.43ms and a median of 309ms.

Hence: The new data seems to be more comparable to Dorrs et al. results (and also to Nyström & Holmqvist 2010 "Eyetracking - A comprehensive guide to methods and measures" books overview (p.23) of eyemovements (duration of saccades in range 200-300ms)) than the old data.

mean fixation durations in seconds for every subject (averaged across runs): ``` 0.3076 0.3264 0.3991 0.3529 0.2128 0.2806 0.2604 0.2797 0.4274 0.3187 0.3758 0.3447 0.3806 0.4073 0.2667 0.2965 0.2932 0.2736 0.3156 0.3413 0.3196 0.3468 0.2721 0.3661 0.3108 0.2803 0.5172 0.2995 0.3465 0.3336 ``` median fixation durations in seconds for every subject (across all runs): ``` 0.2500 0.2490 0.3040 0.2930 0.1705 0.2320 0.2090 0.2350 0.3170 0.2510 0.3060 0.2780 0.3040 0.3220 0.2160 0.2490 0.2360 0.2200 0.2460 0.2630 0.2530 0.2800 0.2200 0.2880 0.2570 0.2320 0.3680 0.2180 0.2620 0.2580 ``` This paper examined eye movements during natural viewing (incl "Hollywood trailers" (different Startwars movies)): Dorr, M., Martinetz, T., Gegenfurtner, K. R. & Barth, E. (2010). Variability of eye movements when viewing dynamic natural scenes. Journal of Vision, 10(10), 1 – 17. (https://jov.arvojournals.org/article.aspx?articleid=2121333) They find: mean: 354ms, median: 253ms. Our overall mean and median (across all subj and runs): overall mean and median 0.326750136304 sec, 0.256 sec. --> really close. (previously, these were the mean and median durations from the old output (the metric is ms): ``` 402.0994 533.5491 616.2017 610.4545 310.0134 500.2420 390.0699 488.8088 512.9236 473.3655 510.7231 374.3638 488.4295 503.1059 474.4354 368.5217 451.8407 412.8786 403.3964 424.4862 425.2159 412.0691 404.4751 437.3975 511.5578 357.6460 633.8693 379.6302 434.9447 401.5308 ``` ``` 286.0000 325.0000 379.0000 416.0000 228.0000 327.0000 277.0000 322.0000 369.0000 323.0000 353.0000 296.0000 352.0000 373.0000 339.0000 283.0000 312.0000 290.0000 280.0000 294.0000 304.0000 303.0000 279.0000 310.0000 365.0000 271.0000 418.0000 254.0000 305.0000 291.0000 ``` which overall amounted to a mean of 444.43ms and a median of 309ms. Hence: The new data seems to be more comparable to Dorrs et al. results (and also to Nyström & Holmqvist 2010 "Eyetracking - A comprehensive guide to methods and measures" books overview (p.23) of eyemovements (duration of saccades in range 200-300ms)) than the old data.

mih commented

2019-04-23 07:13:33 +00:00

(Migrated from github.com)

https://github.com/psychoinformatics-de/paper-remodnav has most of this now. Minus: https://github.com/psychoinformatics-de/remodnav/issues/9