Table 6 Performance comparison of different ensemble approaches on the different complex test datasets.

From: Enhancing occluded and standard bird object recognition using fuzzy-based ensembled computer vision approach with convolutional neural network

Approach

TD1

TD2

TD3

TD4

A

(%)

FS

(%)

A

(%)

FS

(%)

A

(%)

FS

(%)

A

(%)

FS

(%)

Simple Ensemble

Learning

97.27

97.29

97.01

96.89

90.30

91.72

96.96

96.88

Fuzzy-Based

Ensemble

Learning

98.25

98.27

97.97

97.85

90.91

92.28

98.03

98.13

Random Forest

Ensemble

Learning

96.36

96.12

90.00

88.80

90.91

90.81

93.11

93.05

XG-Boost

Ensemble

Learning

80.00

77.65

69.09

66.38

73.33

74.07

87.68

86.91

  1. TD1: Testing Dataset with 25–40% Noise added, TD2: Testing Dataset with 40–50% Noise added, TD3: Test images with random object insertions such as synthetic leaves, shadows, and branches, designed to mimic occlusions in natural settings, TD4: Testing dataset containing completely unseen images downloaded from the internet.
Back to article page