深度学习之基础模型-ShuffleNet

The new architecture utilizes two proposed operations, point wise group convolution and channel shuffle, to greatly reduce computation cost while maintaining accuracy.web 该论文利用了point wise group convol
相关文章
相关标签/搜索