Table 1 Summary of key performance metrics overall accuracy, precision, recall, F1‐score, and Cohen’s κ, for each method against the 823 detections that were located across the ten villages and two cities.

From: Detection of asbestos-based cement rooftops in conflict-affected settings using EnMAP hyperspectral data: a research article

Method

Overall Accuracy (%)

Precision (%)

Recall (%)

F1-score

Kappa

Linear Spectral Unmixing (LSU)

84.5

82.3

86.1

84.1

0.79

Support Vector Machine (SVM)

89.2

87.6

90.3

88.9

0.83

Spectral Angle Mapper (SAM)

86.0

85.2

87.1

86.1

0.81

Adaptive Coherence Estimator

91.4

89.9

92.7

91.2

0.87

Mahalanobis Distance

87.8

86.5

88.9

87.7

0.82

Maximum Likelihood Classification

88.3

86.9

89.5

88.1

0.83

Spectral Information Divergence

90.1

88.8

91.2

90.0

0.85

  1. Metrics are reflective of the final output from each classifier after manual validation through filtering with a requirement of six of eight independent classifier positive matches.
Back to article page