ONNX to TRT Engine: "Assertion failed: mAnchorsCnt == (int) (mAnchorBoxesHost.size() / 4)"
@zerollzeng is already working on this.
Since Dec 7, 2022.
- Dominant language
- C++
- Stars
- 13.4k
- Forks
- 2.4k
- Avg merge
- 5d 3h
- Merged PRs (30d)
- 2
Description
Description
I'm trying to build an engine from an onnx file of a maskrcnn model using the trtexec command but I am getting this assert error.
The exact command I used was:
trtexec --weights mrcnn.onnx --fp16 --verbose
The input dimensions for the model has been changed to (3, 512, 512). I believe this is causing the main issue as the ProposalLayerPlugin is pulling the image shape from "mrcnn_config.h" which has a default input image shape as (3, 1024, 1024). Is there a way to overwrite this to match my new input dimensions?
Environment
I am using a Docker image that pulls nvcr.io/nvidia/tensorrt:22.08-py3
TensorRT Version: 8.4.2.4
NVIDIA GPU: Nvidia GeForce GTX 1050 Ti
NVIDIA Driver Version: 515.43.04
CUDA Version: 11.7.1
CUDNN Version: 8.5.0.9
Operating System: Ubuntu 20.04
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.