1 | initial version |
you mean - theano instead of tensorflow as backend (or, CNTK, even) ?
very unlikely to work. opencv does not know anything about your high-level modeling. and currently, you have to go down to the low-level tf functionality, to export something, opencv's dnn understands
2 | No.2 Revision |
you mean - theano instead of tensorflow as backend (or, CNTK, even) ?
very unlikely to work. opencv does not know anything about your high-level modeling. and currently, you have to go down to the low-level tf functionality, to export something, opencv's dnn understands
(theano seems to "pickle" it's models, and there's no current support to read that in (from c++, too!))