Migrating addConvolution to addConvolutionNd

Hello in the API page addConvolution() is deprecated.

if I am using addConvolutionNd() i get “at least 4 dimensions are required for input”
on the input convolution.
what is the correct way to use the function on a 3 channels input image?

migrating to TensorRT7.1

Hi @gil.gilad,
Please refer to the below link.
https://docs.nvidia.com/deeplearning/tensorrt/api/c_api/classnvinfer1_1_1_i_network_definition.html#a108f9f6d884e642a9e3812b5175fbc55

Thanks!

Thank you for your response, i have seen the API reference and it is still unclear.

If the previous ITensor was an 3X400X400 using the addConvolution(...) function. does the new tensor need to be of size 1X3X400X400 for addConvolutionNd(...)? if so what are the best practiced the squeeze the dimension ?

Hi @gil.gilad,

Are you using --explicitBatch?
If not, request you to share your model and script, so that i can try reproducing it from my end along with the below set of information
o Linux distro and version
o GPU type
o Nvidia driver version
o CUDA version
o CUDNN version
o Python version [if using python]
o Tensorflow and PyTorch version
o TensorRT version
Thanks!