Table 6.

Results for a five-fold cross-validation on the data set: In each row, one of the cameras was declared as the test set, and the images from the four remaining cameras were use for training and validation.

Test Set	PQ	F1	Precision	Recall
Camera 1	0.7849	0.9571	0.9566	0.9575
Camera 2	0.7976	0.9602	0.9616	0.9588
Camera 3	0.7966	0.9513	0.9544	0.9482
Camera 4	0.7653	0.9344	0.9357	0.9331
Camera 5	0.8002	0.9544	0.9540	0.9548
Average	0.7889	0.9515	0.9525	0.9505
STD	0.0129	0.0090	0.0088	0.0094