Training problem - Githubissues

MichalBusta / DeepTextSpotter

285 stars 101 forks source link

Open lfdeep opened 6 years ago

lfdeep commented 6 years ago

If I want to fine-tune the model you gave, then i should put the model you gave in which .py file.

MichalBusta commented 6 years ago

In caffe you need to do:

create slolver: solver = caffe.get_solver('models/tiny_solver.prototxt')
create snapshot: solver.snapshot() this will create .solverstate and .caffemodel
replace *.caffemodel with provided
use it: solver.restore( ...solverstate)