Research Article

[Retracted] Research on Efficient Deep Learning Algorithm Based on ShuffleGhost in the Field of Virtual Reality

Table 1

The comparison of Ghost-like network and three visions of ShuffleGhost.

LayersOutput sizeGhost-like networkShuffleGhostV1ShuffleGhostV2ShuffleGhostV3

conv conv, stride 1 conv, stride 1 conv, stride 1 conv, stride 1
BatchNorm, ReLU
BottleneckGhost bottleneck (Ghost-like module × 2)ShuffleGhost BottleneckV1 (Ghost-like module × 2)ShuffleGhost BottleneckV1 (ShuffleGhost module × 2)ShuffleGhost BottleneckV2 (ShuffleGhost module × 2)
BottleneckGhost bottleneck (Ghost-like module × 2)ShuffleGhost BottleneckV1 (Ghost-like module × 2)ShuffleGhost BottleneckV1 (ShuffleGhost module × 2)ShuffleGhost BottleneckV2 (ShuffleGhost module × 2)
BottleneckGhost bottleneck (Ghost-like module × 2)ShuffleGhost BottleneckV1 (Ghost-like module × 2)ShuffleGhost BottleneckV1 (ShuffleGhost module × 2)ShuffleGhost BottleneckV2 (ShuffleGhost module × 2)
BottleneckGhost bottleneck (Ghost-like module × 2)ShuffleGhost BottleneckV1 (Ghost-like module × 2)ShuffleGhost BottleneckV1 (ShuffleGhost module × 2)ShuffleGhost BottleneckV2 (ShuffleGhost module × 2)
BottleneckGhost bottleneck (Ghost-like module × 2)ShuffleGhost BottleneckV1 (Ghost-like module × 2)ShuffleGhost BottleneckV1 (ShuffleGhost module × 2)ShuffleGhost BottleneckV2 (ShuffleGhost module × 2)
Bottleneck × 3Ghost bottleneck (Ghost-like module × 2)ShuffleGhost BottleneckV1 (Ghost-like module × 2)ShuffleGhost BottleneckV1 (ShuffleGhost module × 2)ShuffleGhost BottleneckV2 (ShuffleGhost module × 2)
BottleneckGhost bottleneck (Ghost-like module × 2)ShuffleGhost BottleneckV1 (Ghost-like module × 2)ShuffleGhost BottleneckV1 (ShuffleGhost module × 2)ShuffleGhost BottleneckV2 (ShuffleGhost module × 2)
Bottleneck × 3Ghost bottleneck (Ghost-like module × 2)ShuffleGhost BottleneckV1 (Ghost-like module × 2)ShuffleGhost BottleneckV1 (ShuffleGhost module × 2)ShuffleGhost BottleneckV2 (ShuffleGhost module × 2)
conv conv, stride 1 conv, stride 1 conv, stride 1 conv, stride 1
BatchNorm, ReLU
PoolingGlobal average poolingGlobal average poolingGlobal average pooling
Dense10D fully connected, softmax10D fully connected, softmax10D fully connected, softmax