Tian, Feng

thanks for raising this bug. I am checking it and will fix asap.

it's quite weird. the code of "bool use_dilation = false;" is in line 92, but your build error shows it's in line 91. are you sure you are using the...

I think you are right. could you contribute a PR for our merge?

@minkkang hi, shall you move GetLocalRate() behind regularization() as the latter also changes the learning_param.diff? it's to be exact same with the no fusion version.

sorry for late respond due to Chinese New Year. Yes, I think your analysis is right. the non-fusion version should be updated.

we didn't formally claim we support opencv4. but as a workaround, you can change CV_BGR2RGB to COLOR_BGR2RGB and have a try.

how about cv::COLOR_BGR2RGB?

Sorry, intelcaffe doesn't support GPU fix point quantization. it's only for CPU.

pls don't touch mkl_xxx layer, it's deprecated.

it's more like out-of-memory allocation issue. how many memory do you have on these two machines?