Tensorrt error


A clear and concise description of the bug or issue.


**TensorRT Version 7.1:
**GPU Type jetson Xavier NX:
**CUDA Version10.2:
**TensorFlow Version (if applicable) 1.15:

Relevant Files

I use slim.conv2d_transpose layver in tensorflow 1.15, my parameters is (inputs,num_outputs=64,kerenl_size=3,stride=2), the input shape is [1,10,10,64], the output shape is [1,20,20,64].
I convert tf model to pb model,then use uff to convert the pb model, however, when I use tensorrt to parse uff model, it comes out a tensorrt error ,the IDeconvolutionLayer outpust shape is [1,60,60,64], it should be [1,20,20,64], I don’t know the reason, how can i fix it.

Hi @lzy_lyx,
UFF parser has been deprecated from TRT 7 onwards,
We recommend you to try pb<< ONNX<< TRT conversion or you can try TF-TRT approach as well.