tensorrt_demos: Problem with SSP custom conversion model

Hello,

Tks for share your work to us. I’m trying to convert a custom YOLOV3 SSP model (2 classes), but I’ve got this error:

` onnx.checker.check_model(yolo_model_def) File “/home/workstation/anaconda3/envs/tensorrtdemo/lib/python3.7/site-packages/onnx/checker.py”, line 93, in check_model C.check_model(model.SerializeToString()) onnx.onnx_cpp2py_export.checker.ValidationError: Op registered for Upsample is deprecated in domain_version of 12

==> Context: Bad node spec: input: “092_convolutional_lrelu” input: “093_upsample_scale” output: “093_upsample” name: “093_upsample” op_type: “Upsample” attribute { name: “mode” s: “nearest” type: STRING }`

Do you have some idea?

About this issue

  • Original URL
  • State: closed
  • Created 4 years ago
  • Comments: 43 (19 by maintainers)

Most upvoted comments

@jkjung-avt and @wjshamblin , it works. Tks! But now, when I’m trying to convert to tensorrt I’ve got the segment fault core dumped error.

@adrianosantospb I concur with wjshamblin. The code in this repository is compatible with “onnx==1.4.1” (opset 10). Your error message shows that you are using opset 12.

So please downgrade your onnx module.

$ sudo pip3 install onnx==1.4.1