Skip to main content

Table 6 Classification evaluation results in fine-tuned model versions pre-trained on the DTD data set and an endoscopic data set using self-supervision

From: Automated classification of polyps using deep learning architectures and few-shot learning

Model

Classification

Acc

Pre

Rec

F1

DTD 

1-nn

75.31 (± 9.41)

75.94 (± 8.63)

75.31 (± 9.41)

75.7 (± 9.8)

 

centroid

81.39 (± 8.53)

82.05 (± 8.61)

81.39 (± 8.53)

81.7 (± 8.4)

 

SVM

81.34 (± 8.74)

81.52 (± 8.39)

81.34 (± 8.74)

81.0 (± 8.6)

self-sv.

1-nn

71.59 (± 8.74)

75.09 (± 8.13)

71.59 (± 8.74)

73.3 (± 9.5)

 

centroid

68.88 (± 8.45)

70.30 (± 8.82)

68.88 (± 8.45)

69.6 (± 9.7)

 

SVM

75.04 (± 8.59)

75.24 (± 8.38)

75.04 (± 8.59)

75.1 (± 8.3)

  1. The table shows the average scores for 100 random training/test splits with \(90\%\) confidence interval