loadModelAndCreateEngine failed


I adapted the sampleUFFMNIST codes for my own application and loaded my .uff file with a base model deploying InceptionV3. However, I get the error:

ERROR: Parameter check failed at: Network.cpp::addInput::287, condition: dims.d[i] > 0
Segmentation fault (core dumped)

when trying to run it. Could someone shed some light as to why this is happening?
Thanks so much!


Do you use any plugin implementation on your model?

Here are some suggestions for your reference:
1. Please check if the input buffer is well-defined.
2. Please check if there is any incorrect layer dimension inside your model.



What kind of plugin implementation are you referring to? I converted my model to a protobuf then to a uff file in python.

Regarding your suggestions:
1a. I would like to check if the input name in parser->registerInput() is the input node name in the original graph (mine is “prefix/input_1”)?

1b. The input type of my original is float32, will this be considered as kFLOAT for the type?

1c. The dimensions I used (DimsCHW(1, 256, 1024)) seem to be correct, though the original model takes in an input of shape (?, 256, 1024, 1)

  1. The model runs well on normal keras python code so i don’t think there is any incorrect layer dimension.

Thanks so much for your help in this!

I fixed the input issue. I registered the wrong input node name with the parser. Thanks.