Table 4.
The accuracy (%) of the CNN models with image size = 32. The Accuracy Test Set was used for accuracy test for automatic identification of the five Korean softwoods
Species | LeNet | LeNet2 | LeNet3 | MiniVGGNet | MiniVGGNet2 | MiniVGGNet3 |
Cedar | 97.1 | 85.4 | 99.5 | 99.9 | 99.7 | 100.0 |
Cypress | 51.4 | 83.9 | 79.2 | 25.4 | 79.0 | 62.0 |
Korean Pine | 27.3 | 46.7 | 59.7 | 20.6 | 49.2 | 35.7 |
Korean Red Pine | 85.3 | 81.7 | 84.4 | 35.1 | 40.3 | 35.6 |
Larch | 99.3 | 99.7 | 98.2 | 25.0 | 38.0 | 43.2 |
Average | 72.1 | 79.5 | 84.2 | 41.2 | 61.2 | 55.3 |
Standard Deviation | 28.2 | 17.6 | 14.5 | 29.7 | 24.2 | 24.3 |