Research Article
Remote Sensing Landslide Recognition Based on Convolutional Neural Network
Table 3
Results of the comparison between CNN, SVM, GBM, and NBC.
| Evaluation parameter | CNN | SVM | GBM | NBC | 1 band (%) | RGB (%) | 4 bands + NDVI (%) | 1 band (%) | RGB (%) | 4 bands + NDVI (%) | 1 band (%) | RGB (%) | 4 bands + NDVI (%) | 1 band (%) | RGB (%) | 4 bands + NDVI (%) |
| Accuracy | 89.02 | 98.52 | 98.98 | 88.43 | 98.48 | 98.83 | 87.67 | 98.32 | 98.85 | 60.40 | 83.20 | 97.84 | Precision | 20.9 | 91.64 | 93.82 | 30.54 | 93.23 | 94.66 | 37.71 | 92.22 | 94.72 | 82.66 | 89.50 | 95.04 | Recall | 51.91 | 94.95 | 96.89 | 47.13 | 93.19 | 94.85 | 43.85 | 92.74 | 94.99 | 19.66 | 38.98 | 83.41 | F-measure | 29.80 | 93.27 | 95.33 | 37.06 | 93.21 | 94.75 | 40.55 | 92.485 | 94.85 | 31.76 | 54.31 | 88.85 |
|
|