Table 5.

Runtime evaluation of the proposed experiments for categorical segmentation and combined instance segmentation: All values are given as mean runtime and standard deviation over the 226 images of the test set.

Categorical	Network Inference (ms)	Postprocessing (ms)
ResNet34	$32.47\pm 0.31$	$2.47\pm 0.22$
Incep.-RN-v2	$88.91\pm 0.23$
Combined	Network Inference (ms)	Postprocessing (incl. Clustering) (ms)	Clustering (ms)
ResNet34	$41.98\pm 1.12$	$2115.73\pm 448.26$	$2068.53\pm 443.72$
Incep.-RN-v2	$101.39\pm 2.41$