Table 8.
Performance comparison of LCAMNet and several classical CNN models on the FGVC8 dataset.
| Models | Accuracy (%) | Recall (%) | Precision (%) | F1 (%) | Kappa (%) | MCC (%) | GFLOPs (G) | Parameters (M) |
|---|---|---|---|---|---|---|---|---|
| GoogLeNet | 95.31 | 95.33 | 95.56 | 95.39 | 94.20 | 95.39 | 1.60 | 7.01 |
| InceptionRes-NetV2 | 94.92 | 94.93 | 95.40 | 95.06 | 93.71 | 95.06 | 6.50 | 55.84 |
| EfficientNet-B0 | 91.80 | 91.88 | 92.13 | 91.92 | 89.76 | 91.92 | 0.42 | 8.43 |
| ShuffleNetV2 | 91.80 | 91.96 | 91.89 | 91.86 | 89.91 | 91.86 | 0.04 | 1.31 |
| LCAMNet | 95.31 | 95.29 | 95.81 | 95.39 | 94.22 | 95.39 | 0.03 | 1.30 |
Bold values indicate the best results in each column. For Accuracy, Recall, Precision, F1, Kappa, and Matthews Correlation Coefficient (MCC), higher values indicate better performance. Conversely, for GFLOPs and Parameters, lower values indicate a more lightweight and efficient model.