solved
we are reviewing now and will keep you updated.
can you share the full runtime output?
Sorry for the late update. Below is the runtime output. The model could run in caffe correctly.
####### input args#######
C=3;H=608;W=608;caffemodel=./mobilenet_deconv_maxpool.caffemodel;calib=;class=1;mode=fp32;outputs=conv25,conv26,conv27;prototxt=./mobilenet_deconv_maxpool.prototxt;
init plugin proto: ./mobilenet_deconv_maxpool.prototxt caffemodel: ./mobilenet_deconv_maxpool.caffemodel
Begin parsing model…
Input:1024 13 13
ouputDims:1024 26 26
ERROR: conv17/sum: all concat input tensors must have the same dimensions except on the concatenation axis
run3: helpers.cpp:56: nvinfer1::DimsCHW nvinfer1::getCHW(const nvinfer1::Dims&): Assertion `d.nbDims >= 3’ failed.
Aborted (core dumped)
I have tried to remove the maxpooling layer and connect the upsample layer with the conv15_e/dw(modified the param in upsample layer to make sure the output width and height same with conv17), it works fine(without consider the weights, only prototxt). So the problem may caused by the maxpooling layer I guess.
Hello,
to help us debug, can you share a small repro containing the parsing code and model that demonstrate the error you are seeing? You can DM me the files.