Table 2 Comparison of classification performance and computational complexity among different models.
Model | A% | P% | R% | F1% | Params (K) | FLOPs(M) | Memory (MB) | Time (min) |
|---|---|---|---|---|---|---|---|---|
SVM | 81.32 | 85.32 | 81.31 | 80.31 | 11.2 | 4.6 | 15.2 | 2.4 |
XGB | 84.21 | 90.73 | 84.26 | 84.21 | 72.5 | 38.4 | 28.7 | 4.5 |
BPNN | 86.45 | 87.59 | 86.65 | 86.36 | 222.4 | 67.1 | 45.8 | 9.1 |
ResNet | 87.59 | 88.73 | 88.84 | 88.47 | 8124 | 1920 | 487.3 | 79 |
WCNN | 88.73 | 91.18 | 89.40 | 89.29 | 3024 | 920 | 248.7 | 61 |
MobileNet | 90.94 | 92 | 91.23 | 91.50 | 1632 | 450 | 156.9 | 46 |
DenseNet | 92.14 | 93.11 | 92.26 | 92.30 | 5021 | 1800 | 542.1 | 66 |
Swin-T | 93.25 | 93.87 | 93.58 | 93.24 | 2467 | 6500 | 1265.8 | 94 |
ARDDMA-Net | 98.36 | 98.21 | 98.48 | 97.76 | 2303 | 860 | 213.5 | 55 |
