Advances, Systems and Applications
Table 5 The performance of some RDPNet models on the ImageNet dataset
From: RDPNet: a single-path lightweight CNN with re-parameterization for CPU-type edge devices
Model | Top-1 Acc.(\(\%\)) | Top-5 Acc.(\(\%\)) | Latency(ms) |
|---|---|---|---|
MobileNetV2-0.5x | 65.03 | 85.72 | 38.52 |
MobileNetV3-large-0.35x | 64.78 | 83.56 | 23.83 |
ShuffleNetV2-0.5x | 61.32 | 83.73 | 25.58 |
RDPNet-0.75x | 67.23 | 88.93 | 18.58 |
MobileNetV1-1x | 70.99 | 89.68 | 40.76 |
MobileNetV2-1x | 72.15 | 90.65 | 46.62 |
MobileNetV3-small-1.25x | 70.67 | 89.51 | 37.08 |
ShuffleNetV2-1.5x | 71.63 | 90.15 | 36.76 |
RDPNet-1.25x | 73.65 | 91.82 | 23.60 |