Skip to main content

Table 4 Classification of models trained on grayscale images

From: Activity landscape image analysis using convolutional neural networks

CollectionRFSVMCNNMetric
10.57 ± 0.010.53 ± 0.010.71 ± 0.02Accuracy
0.57 ± 0.010.54 ± 0.010.71 ± 0.02F1
0.35 ± 0.010.30 ± 0.010.56 ± 0.03MCC
20.54 ± 0.010.53 ± 0.010.70 ± 0.03Accuracy
0.55 ± 0.010.54 ± 0.010.70 ± 0.03F1
0.32 ± 0.020.29 ± 0.020.55 ± 0.04MCC
30.55 ± 0.020.53 ± 0.010.70 ± 0.03Accuracy
0.56 ± 0.010.54 ± 0.020.70 ± 0.03F1
0.33 ± 0.020.30 ± 0.020.55 ± 0.04MCC
40.54 ± 0.020.57 ± 0.030.67 ± 0.03Accuracy
0.54 ± 0.020.58 ± 0.040.67 ± 0.03F1
0.31 ± 0.030.36 ± 0.050.51 ± 0.05MCC
50.55 ± 0.030.50 ± 0.010.68 ± 0.02Accuracy
0.56 ± 0.030.51 ± 0.020.68 ± 0.02F1
0.33 ± 0.040.25 ± 0.020.52 ± 0.03MCC
60.58 ± 0.010.53 ± 0.020.72 ± 0.03Accuracy
0.58 ± 0.020.55 ± 0.020.72 ± 0.03F1
0.37 ± 0.020.30 ± 0.030.59 ± 0.04MCC
70.69 ± 0.020.68 ± 0.010.74 ± 0.04Accuracy
0.69 ± 0.020.68 ± 0.010.74 ± 0.04F1
0.53 ± 0.030.52 ± 0.020.62 ± 0.06MCC
  1. The table summarizes classification performance for color-coded 3D AL and Ref-AL images using RF, SVM, and CNN models trained on grayscale images. All values reported are averages and standard deviations over 10 independent trials