感谢。侵删。
参考自:https://mp.weixin.qq.com/s?__biz=MzIwOTc2MTUyMg==&mid=2247491298&idx=2&sn=3927e433c4b37cc8ccf06561def4eb8c&chksm=976fbb7fa01832692196acc23c6012b6c33a14e4d7804ae69ade08d0fa45552bf240290a1002&mpshare=1&scene=1&srcid=&sharer_sharetime=1566490799501&sharer_shareid=7955453dfe9f1a47a0cff7902ea03b0b&key=a6bb304d72cb772f900ffdfbddd758c78d6560bdf6ea053f174cc09606d3ca2ff17022084e3a1a17f4af6cc35ddcd3a6848d90393ae97bf37d51a2e56aef331ddfddc0ccb9da710f0a37fcb4c76e7aed&ascene=1&uin=ODY2NTI4OTYz&devicetype=Windows+10&version=62060833&lang=zh_CN&pass_ticket=R5ri6brIVXX5nbg6iwFTAU7ZO4kLuMRUYLxCOMtVk5y4j86UCUvgBscIT%2BWJs%2BLq
NN-SVG
画图工具地址:http://alexlenail.me/NN-SVG/
只能画卷积,短接层好像画不了,复杂结构画不了
PlotNeuralNet
项目地址:https://github.com/HarisIqbal88/PlotNeuralNet
颜值高,能画复杂结构,但需要用latex写
类似工具:https://github.com/jettan/tikz_cnn
Draw_Convnet
项目地址:https://github.com/gwding/draw_convnet
纯python画图,图片比较严肃,适合写论文
类似工具:https://github.com/yu4u/convnet-drawer
Netscope
项目地址:https://github.com/ethereon/netscope
caffe网络可视化工具。在不需要生动体现网络结构时,只需要使用方框表示层和连接关系时,用这个比较合适