panchengl

yes, i met same problem, i think retrained mode will solve this problem？@sunnyden

 Thank you very much for your reply. I modified the resolution in the export code（640x604->320x240）. but i met the same phenomenon, Do you mean I need to modify other...

ok, I will test and reply to you tomorrow morning， thank U

hello, i change 640x640 -> 320x320, but get the same error, onnx create process like this:  wrong boxes like this:  @sunnyden

hey， how do u solve problem, Modify concat to conv, and then retrain the network?@kadebudongtan

yes, u are right, problem solved, thanks@Skr20200701

hey, thanks your work, i use youer code class Focus to replace oprignal Focus: first phenomenon: model can be run and don`t modify any others. second phenomenon: map was declined....

@kadebudongtan

Hello, this work has been done for a long time. I forgot some parameters, but obviously, cuda memory will be reduced a lot, especially in centernet_model train and inference. After...

This version only support resnet101, dcn and no_dcn is ok, dcn modue added at backbone last in centernet, dcn does not affect pruning