Table 1 ShuffleNetV2 network structure.
From: Micro-expression recognition model based on TV-L1 optical flow method and improved ShuffleNet
Layers | Output size | Kernel Size | Stride | Repeat | Output channels |
|---|---|---|---|---|---|
Image | 224 × 224 | – | – | – | 3 |
Conv1 | 112 × 112 | 3 × 3 | 2 | 1 | 24 |
MaxPool | 56 × 56 | 3 × 3 | 2 |  |  |
Stage 2 | 28 × 28 | 1 × 1 | 2 | 1 | 116 |
|  | 28 × 28 | 1 × 1 | 1 | 3 |  |
Stage 3 | 14 × 14 | 1 × 1 | 2 | 1 | 232 |
|  | 14 × 14 | 1 × 1 | 1 | 7 |  |
Stage 4 | 7 × 7 | 1 × 1 | 2 | 1 | 464 |
|  | 7 × 7 | 1 × 1 | 1 | 3 |  |
Conv 5 | 7 × 7 | 1 × 1 | 1 | 1 | 1024 |
GlobalPool | 1 × 1 | 7 × 7 | – | – | – |
