Table 3 Detection performance of different models on different validation datasets.

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

Model

Cascade Mask R-CNN

YOLACT

SOLOv2

DetectoRS

Mask Deformable DETR

Adapt-Cascade

bboxmAP/%

segm mAP/%

bbox mAP/%

segm mAP/%

bbox mAP/%

segm mAP/%

bbox mAP/%

segm mAP/%

bbox mAP/%

segm mAP/%

bbox mAP/%

segm mAP/%

Validation set

83.35

81.18

79.44

77.02

81.32

79.47

84.05

82.72

82.94

80.88

91.06

89.68

Daytime static group

87.02

84.63

82.83

80.41

84.94

82.96

87.72

86.12

86.77

84.24

92.45

91.02

Daytime moving group

80.17

78.65

76.38

74.84

78.34

77.01

80.93

80.12

79.72

78.55

90.87

88.46

Daytime mixed group

81.89

79.65

78.12

75.43

80.14

78.02

82.78

81.13

81.63

79.35

90.92

89.27

Nighttime static group

85.73

83.81

81.54

79.76

83.56

82.13

86.45

85.58

85.42

83.48

92.17

90.82

Nighttime moving group

78.29

76.53

74.71

72.58

76.57

74.88

79.03

77.92

78.06

76.44

90.42

88.96

Nighttime mixed group

78.86

76.95

75.19

73.14

77.64

75.31

79.61

78.36

78.42

76.64

90.76

89.26