Bidimensional regression can be used to transform the eye gaze data into a the screen coordinate system using a calibration sequence. For this, we use known target coordinates as independent variables. Please note that the example below assumes that participants fixate faithfully for most of the time and that recording artifacts, such as blinks, were already removed. This example will use the example dataset.
ggplot(data= EyegazeData, aes(x= x, y= y, color= target, fill= target)) +
geom_point(data= EyegazeData %>% group_by(target, target_x, target_y) %>% summarise(.groups="drop"),
aes(x= target_x, y= target_y), shape= 21, size= 10, fill= 'white') +
geom_point(alpha= 0.5, shape= 21, show.legend=FALSE) +
ggtitle('Raw eye gaze')
<- fit_transformation(target_x + target_y ~ x + y, EyegazeData, transformation = 'affine')
lm2aff #>
#> SAMPLING FOR MODEL 'tridim_transformation' NOW (CHAIN 1).
#> Chain 1:
#> Chain 1: Gradient evaluation took 0.001 seconds
#> Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 10 seconds.
#> Chain 1: Adjust your expectations accordingly!
#> Chain 1:
#> Chain 1:
#> Chain 1: Iteration: 1 / 2000 [ 0%] (Warmup)
#> Chain 1: Iteration: 200 / 2000 [ 10%] (Warmup)
#> Chain 1: Iteration: 400 / 2000 [ 20%] (Warmup)
#> Chain 1: Iteration: 600 / 2000 [ 30%] (Warmup)
#> Chain 1: Iteration: 800 / 2000 [ 40%] (Warmup)
#> Chain 1: Iteration: 1000 / 2000 [ 50%] (Warmup)
#> Chain 1: Iteration: 1001 / 2000 [ 50%] (Sampling)
#> Chain 1: Iteration: 1200 / 2000 [ 60%] (Sampling)
#> Chain 1: Iteration: 1400 / 2000 [ 70%] (Sampling)
#> Chain 1: Iteration: 1600 / 2000 [ 80%] (Sampling)
#> Chain 1: Iteration: 1800 / 2000 [ 90%] (Sampling)
#> Chain 1: Iteration: 2000 / 2000 [100%] (Sampling)
#> Chain 1:
#> Chain 1: Elapsed Time: 11.729 seconds (Warm-up)
#> Chain 1: 6.247 seconds (Sampling)
#> Chain 1: 17.976 seconds (Total)
#> Chain 1:
<- predict(lm2aff, summary=TRUE, probs=NULL)
adjusted_gaze colnames(adjusted_gaze) <- c('adjX', 'adjY')
<- cbind(EyegazeData, adjusted_gaze)
adjusted_gaze
ggplot(data= adjusted_gaze, aes(x= adjX, y= adjY, color= target, fill= target)) +
geom_point(data= adjusted_gaze %>% group_by(target, target_x, target_y) %>% summarise(.groups="drop"),
aes(x= target_x, y= target_y), shape= 21, size= 10, fill= 'white') +
geom_point(alpha= 0.5, shape = 21, show.legend = FALSE) +
xlab('x')+
ylab('y')+
ggtitle('Adjusted eye gaze')