Table 2.

Ranking of detection task of polyp generalisation challenge.

Team/method	Avg_det $\uparrow$	Avg. deviation scores $\downarrow$				Time $\downarrow$	Rank $\downarrow$
		dev_g1-3	dev_g2-3	dev_g4-3	dev_g	(in ms)
AIM_CityU18	0.450	0.089	0.134	0.056	0.093	100	1
GECE_VISION20	0.384	0.056	0.253	0.069	0.126	320	5
HoLLYS_ETRI24	0.491	0.122	0.212	0.098	0.144	690	2
JIN_ZJU19	0.478	0.062	0.230	0.091	0.128	1900	3
YOLOv427	0.316	0.099	0.178	0.060	0.112	13	6
RetinaNet (ResNet50)28	0.320	0.031	0.086	0.040	0.052	27	4
EfficientDetD229	0.298	0.058	0.173	0.078	0.103	200	7

Average precision across all test splits is provided as Avg_det. Deviation scores are calculated between the test data 3 w.r.t. data 1 (dev_g_1-3), data 2 (dev_g_2-3) and data 4 (dev_g_4-3). An average deviation score dev_g is computed by averaging the computed deviations for each data. Test execution time is provided in ms. Finally, a rank column is used to provide an average rank based on the computed ranks for each Avg_det, dev_g and time. Top-two values for each metric are highlighted in bold.

$\uparrow$: best increasing         $\downarrow$: best decreasing.