How can I get the whole Network structure of FAN ?

1adrianb / face-alignment-training

Training code for the networks described in "How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)" paper.

https://www.adrianbulat.com/face-alignment

Other

145 stars 48 forks source link

How can I get the whole Network structure of FAN ? #22

Open colacool opened 6 years ago

colacool commented 6 years ago

I have tried your python code of FAN which actually has a very good result. But I have a visualize problem. Sorry that i am unfamiliar with lua and torch. I have tried python-graphviz in your project in FAN in python but I failed with getting "NoneType" pdf which contained nothing. Could you give a whole visualized network structure with some specific values like netscope of Caffe or Tensorboard of tensorflow? I mean the network of 2D face feature structure is OK, all structure is best!

WaterCube001 commented 5 years ago

Hi, I used: graph.dot(model.fg, 'MODEL', 'myMODEL') 　in the ./modelfan.lua to visualize the fan model. I get the architecture graph(.svg format) of fan. However, it is a huge one and hard to capture as a explicit image. By the way, for .svg format, it must be opened in a browser. Thx.