pytorch ShuffleNetV2 知识整理

产品设计

 

 

算法提出是从这里引出的：

https://blog.csdn.net/u012426298/article/details/80688080

 

各种算法比较：

https://blog.csdn.net/sun_28/article/details/78170878

 

pytorch实现，有torch转caffe：

https://github.com/miaow1988/ShuffleNet_V2_pytorch_caffe

https://github.com/Randl/ShuffleNetV2-pytorch

这个有测试速度，cpu 416 100ms

改变通道数：1:改为了96通道后，速度变成了200ms了，与mobilenetv2就差不多了

self.num_of_channels = {0.5: [24, 48, 96, 192, 1024], 1: [96, 116, 232, 464, 1024],
                        1.5: [24, 176, 352, 704, 1024], 2: [24, 244, 488, 976, 2048]}

https://github.com/jacke121/ShuffleNetV2-pytorch

https://github.com/timctho/shufflenet-v2-pytorch/blob/master/net.py

这有测试

https://github.com/ZHANG-SHI-CHANG/ShuffleNetV2-Pytorch/blob/master/ShuffleNetV2.py

https://github.com/xingmimfl/pytorch_ShuffleNet_ShuffleNetV2/blob/master/shufflenet_v2.py

https://github.com/Bugdragon/ShuffleNet_v2_PyTorch/blob/master/ShuffleNetv2.py
https://github.com/Randl/ShuffleNetV2-pytorch