Models | Params (Million) | Reduction | FLOPs (Billion) | Reduction | Accuracy |
Res2Net | 23.53 | - | 1.75 | - | 0.788 |
ShuffleNetV2 | 1.13 | - | 1.24 | - | 0.755 |
MnasNet | 3.12 | - | 0.93 | - | 0.611 |
MobileNetV3 | 2.24 | - | 0.73 | - | 0.731 |
MobilenetV1 | 3.21 | 0% | 0.567 | 0% | 0.926 |
CMSFL-Net | 0.86 | - | 0.65 | - | 0.794 |
kMobileNet16Ch | 0.244 | 96.40% | 0.092 | 83.79% | 0.885 |
kMobileNet32Ch | 0.403 | 87.47% | 0.153 | 72.90% | 0.910 |
kMobileNet64Ch | 0.718 | 77.66% | 0.251 | 55.65% | 0.920 |
kMobileNet128Ch | 1.32 | 58.83% | 0.370 | 34.76% | 0.923 |
HMFF-MobileNet16Ch | 0.245 | 92.35% | 0.090 | 84.11% | 0.898 |
HMFF-MobileNet32Ch | 0.375 | 88.31% | 0.148 | 73.89% | 0.913 |
HMFF-MobileNet64Ch | 0.633 | 80.29% | 0.238 | 57.98% | 0.921 |
HMFF-MobileNet128Ch | 1.12 | 65.02% | 0.341 | 39.78% | 0.927 |