Research Article
[Retracted] Research on Efficient Deep Learning Algorithm Based on ShuffleGhost in the Field of Virtual Reality
Table 2
The structure of Ghost-like network for tiny ResNet experiment.
| | Layers | Output size | Ghost-like network | Parameters |
| | Convolution | | conv, stride 1 | 216 | | BatchNorm, ReLU | | | 6 | | Ghost bottleneck | | Ghost-like module × 2 | 3604 | | Ghost bottleneck | | Ghost-like module × 2 | 13800 | | Ghost bottleneck | | Ghost-like module × 2 | 34680 | | Ghost bottleneck | | Ghost-like module × 2 | 61214 | | Ghost bottleneck | | Ghost-like module × 2 | 116432 | | Ghost bottleneck × 3 | | Ghost-like module × 2 | 218960, 174880, 174880 | | Ghost bottleneck | | Ghost-like module × 2 | 276592 | | Ghost bottleneck × 3 | | Ghost-like module × 2 | 718288, 695360, 695360 | | Convolution | | conv, stride 1 | 4800 | | BatchNorm, ReLU | | | 960 | | Pooling | | Global average pooling | 0 | | Dense | | 10D fully connected, softmax | 76810 | | | | 3266852 |
|
|