overlay_agreement() answers the same three-way question about pixels, and on an instance result that is the wrong unit: a merged pair and a split object are both nearly perfect pixelwise and both wrong, which is exactly what an average precision will not tell anybody. This matches whole objects at one IoU threshold, the way the metric matches them, so the picture cannot disagree with the number it illustrates.
overlay_instance_agreement(
image,
prediction,
truth,
threshold = 0.5,
alpha = drawing_style()\$overlay_alpha,
colours = drawing_style()\$agreement_colours
)
image
|
One image, as overlay_instances() takes it.
|
prediction
|
An instance label map from a model. |
truth
|
An instance label map from the annotation. |
threshold
|
The IoU above which a predicted object counts as a true one, strictly greater. 0.5 is the loosest of the challenge’s ten thresholds, so it is the kindest reading, and below 0.5 is refused for the reason the metric refuses it: there one prediction can satisfy two true objects and the match stops being one to one. |
alpha
|
How much of the tint shows. |
colours
|
Named hit, missed and false_alarm, as overlay_agreement() takes them.
|
A height x width x 3 array of integers, 0-255.
overlay_instances() for the objects themselves.
library("rplatypus")
image <- array(0.5, dim = c(5, 11, 3))
truth <- matrix(0L, nrow = 5, ncol = 11)
truth[2:4, 2:4] <- 1L
truth[2:4, 6:8] <- 2L
predicted <- matrix(0L, nrow = 5, ncol = 11)
predicted[2:4, 2:4] <- 1L # one found
predicted[2:4, 9:11] <- 2L # one invented, and object 2 missed
shown <- overlay_instance_agreement(image, predicted, truth)
# Three different colours: hit, missed, invented.
nrow(unique(rbind(shown[3, 2, ], shown[3, 6, ], shown[3, 10, ])))[1] 3