VisualComputingInstitute
for eval
very hopeful for your help：
i use your fastercnn.py as a moudle in my model , when train it is very good the result is good too
but when eval i just change the towersetup the istrain is false , but there is no output box mask
i try to find the reason, but i cant could you please teach me is there something else i need to do when i use the fasterrcnn to eval , so it can output box prob mask
it seem everthing is normal before the create heaeds
it seem like there is somethings wrong in _create_fastrcnn_output
when istrain is faslse
when istrain true
Hi xuchen,
note that there is no need to change the arguments to tower_setup. Instead, please change the config (create a separate config file for eval, or override using the command line). The easiest should be to see how we do "forwarding and tracking" in the README.
Best,
Paul
