How to adjust parameters for resnet101 backbone ?

CasonTsai / MaskRcnn_tensorflow_cpp_inference

inference mask_rcnn model with tensorflow c++ api

MIT License

46 stars 10 forks source link

How to adjust parameters for resnet101 backbone ? #3

Open jediofgever opened 4 years ago

jediofgever commented 4 years ago

Hi, thank you for the great work, I am trying out this project under linux, I am trying to make the ballon detector run in C++,
I compiled the code without errors but when I run the code I encounter the following problem ,

Assertion `dimensions_match(m_leftImpl.dimensions(), m_rightImpl.dimensions())' failed.
Aborted

I think I am missing some parameter adjustment for my use case, my configuration is the one in balloon.py of maskrcnn example. Do you know how can I overcome this ?

CasonTsai commented 4 years ago

@jediofgever i am sorry ,it's my negligence that i ignored some details.Some guys has met same problem with you.the error may be caused by the difference between input of model and size of image ，which i dont described clearly in that code.but i am ouside at now time,I will find a time to explain clearly,could you provide more error details such as the input or config of your network and the size of image when you inferenced the model?