Skip to main content

Table 1 Data is shown regarding the accuracy rates of both baseline and ad hoc trained methods.

From: Scoring pleurisy in slaughtered pigs using convolutional neural networks

Method

Accuracy rates (%)

Healthy half carcasses (class I)

Lesions class II

Lesions class III

Lesions class IV

Average value

Baseline DL method

92

70

62

68

73

Ad hoc trained DL-based method

96

70

84

92

85.5

  1. Accuracy values clearly indicate that the segmentation-based approach provides better results, the only exception being represented by class II.