Table 2 Empirical ablation study results on LDPolypVideo dataset.

From: Self-supervised representation learning using feature pyramid siamese networks for colorectal polyp detection

Method

Weight \(\lambda \)

mAP

AP\(_{50}\)

AP\(_{75}\)

Base

N/A

14.3

35.4

6.1

Base+FP

0.5

16.7 (+ 2.4)

42.5

5.9

Base+img-pre

N/A

23.0 (+ 8.7)

47.3

19.6

FPSiam (ours)

0.0

22.1 (+ 7.8)

46.7

17.5

0.25

23.2 (+ 8.9)

48.0

19.3

0.5

24.3 (+ 10.0)

49.6

20.5

0.75

23.5 (+ 9.2)

47.7

19.8

1.0

22.9 (+ 8.6)

46.0

20.0

  1. Significant values are in [bold].
Back to article page