Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update README.md, deploy model with tensorrt #623

Merged
merged 2 commits into from
Aug 4, 2020

Conversation

wang-xinyu
Copy link
Contributor

@wang-xinyu wang-xinyu commented Aug 4, 2020

πŸ› οΈ PR Summary

Made with ❀️ by Ultralytics Actions

🌟 Summary

Added TensorRT Deployment reference to the README.

πŸ“Š Key Changes

  • Included a new link in the README to a TensorRT Deployment resource.

🎯 Purpose & Impact

  • πŸš€ Purpose: Provides users with a reference to TensorRT Deployment, likely to assist with more efficient model deployment on NVIDIA GPUs.
  • πŸ’‘ Impact: Users interested in optimizing their YOLOv5 models for production may benefit from easier access to relevant information on deploying with TensorRT, potentially leading to enhanced performance and faster inference times.

@glenn-jocher glenn-jocher merged commit 7ecf09d into ultralytics:master Aug 4, 2020
@glenn-jocher
Copy link
Member

glenn-jocher commented Aug 4, 2020

Great, thanks! I've updated the wiki and the export tutorial also at https://docs.ultralytics.com/yolov5/tutorials/model_export

@wang-xinyu
Copy link
Contributor Author

@glenn-jocher You are welcome. Actually we were not using ONNX, we define the yolov5-s/m/l/x network directly with TensorRT API.

We convert the .pt weight file to .wts(plain text), and then load .wts into tensorrt and build tensorrt model.

this is my repo intro

TensorRTx

TensorRTx aims to implement popular deep learning networks with tensorrt network definition APIs. As we know, tensorrt has builtin parsers, including caffeparser, uffparser, onnxparser, etc. But when we use these parsers, we often run into some "unsupported operations or layers" problems, especially some state-of-the-art models are using new type of layers.

So why don't we just skip all parsers? We just use TensorRT network definition APIs to build the whole network, it's not so complicated.

I wrote this project to get familiar with tensorrt API, and also to share and learn from the community.

All the models are implemented in pytorch or mxnet first, and export a weights file xxx.wts, and then use tensorrt to load weights, define network and do inference. Some pytorch implementations can be found in my repo Pytorchx, the remaining are from polular open-source implementations.

@glenn-jocher
Copy link
Member

@wang-xinyu oh wow ok! Did not know that, I will correct the export tutorial.

KMint1819 pushed a commit to KMint1819/yolov5 that referenced this pull request May 12, 2021
* Update README.md, deploy model with tensorrt

* Update README.md

Co-authored-by: Glenn Jocher <[email protected]>
BjarneKuehl pushed a commit to fhkiel-mlaip/yolov5 that referenced this pull request Aug 26, 2022
* Update README.md, deploy model with tensorrt

* Update README.md

Co-authored-by: Glenn Jocher <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants