Table 2 The optimal values of Bbox mAP, Segm mAP and loss during the training process.

From: Motion blur aware multiscale adaptive cascade framework for ear tag dropout detection in reserve breeding pigs

Model

Bbox mAP/%

Segm mAP/%

Recall/%

FPS

Cascade Mask R-CNN

83.35

81.18

89.69

21.64

YOLACT

79.44

77.02

85.98

23.18

SOLOv2

81.32

79.47

88.64

20.61

DetectoRS

84.05

82.72

90.31

17.53

Mask Deformable DETR

82.94

80.88

88.96

18.94

Adapt-Cascade

91.06

89.68

96.32

19.20

Back to article page