Hiwyl / yolov5-rt-stack

yolort is a runtime stack for yolov5 on specialized accelerators such as libtorch, onnxruntime, tvm, tensorrt and ncnn.

YOLOv5 Runtime Stack

---

Documentation • Installation Instructions • Deployment • Contributing • Reporting Issues

---

---

🤗 Introduction

What it is. Yet another implementation of Ultralytics's YOLOv5. yolort aims to make the training and inference of the object detection integrate more seamlessly together. yolort now adopts the same model structure as the official YOLOv5. The significant difference is that we adopt the dynamic shape mechanism, and within this, we can embed both pre-processing (letterbox) and post-processing (nms) into the model graph, which simplifies the deployment strategy. In this sense, yolort makes it possible to be deployed more friendly on LibTorch, ONNXRuntime, TVM and so on.

About the code. Follow the design principle of detr:

object detection should not be more difficult than classification, and should not require complex libraries for training and inference.

yolort is very simple to implement and experiment with. You like the implementation of torchvision's faster-rcnn, retinanet or detr? You like yolov5? You love yolort!

🆕 What's New

• Dec. 27, 2021. Add TensorRT C++ interface example. Thanks to Shiquan.
• Dec. 25, 2021. Support exporting to TensorRT, and inferencing with TensorRT Python interface.
• Sep. 24, 2021. Add ONNXRuntime C++ interface example. Thanks to Fidan.
• Feb. 5, 2021. Add TVM compile and inference notebooks.
• Nov. 21, 2020. Add graph visualization tools.
• Nov. 17, 2020. Support exporting to ONNX, and inferencing with ONNXRuntime Python interface.
• Nov. 16, 2020. Refactor YOLO modules and support dynamic shape/batch inference.
• Nov. 4, 2020. Add LibTorch C++ inference example.
• Oct. 8, 2020. Support exporting to TorchScript model.

🛠️ Usage

There are no extra compiled components in yolort and package dependencies are minimal, so the code is very simple to use.

Installation and Inference Examples

• Above all, follow the official instructions to install PyTorch 1.7.0+ and torchvision 0.8.1+

• Installation via Pip

  Simple installation from PyPI

  pip install -U yolort

  Or from Source

  # clone yolort repository locally
  git clone https://github.com/zhiqwang/yolov5-rt-stack.git
  cd yolov5-rt-stack
  # install in editable mode
  pip install -e .

• Install pycocotools (for evaluation on COCO):

  pip install -U 'git+https://github.com/ppwwyyxx/cocoapi.git#subdirectory=PythonAPI'

• To read a source of image(s) and detect its objects 🔥

  from yolort.models import yolov5s
  
  # Load model
  model = yolov5s(pretrained=True, score_thresh=0.45)
  model.eval()
  
  # Perform inference on an image file
  predictions = model.predict("bus.jpg")
  # Perform inference on a list of image files
  predictions = model.predict(["bus.jpg", "zidane.jpg"])

Loading via torch.hub

The models are also available via torch hub, to load yolov5s with pretrained weights simply do:

model = torch.hub.load("zhiqwang/yolov5-rt-stack:main", "yolov5s", pretrained=True)

Loading checkpoint from official yolov5

The following is the interface for loading the checkpoint weights trained with ultralytics/yolov5. See our how-to-align-with-ultralytics-yolov5 notebook for more details.

from yolort.models import YOLOv5

# 'yolov5s.pt' is downloaded from https://github.com/ultralytics/yolov5/releases/download/v5.0/yolov5s.pt
ckpt_path_from_ultralytics = "yolov5s.pt"
model = YOLOv5.load_from_yolov5(ckpt_path_from_ultralytics, score_thresh=0.25)

model.eval()
img_path = "test/assets/bus.jpg"
predictions = model.predict(img_path)

🚀 Deployment

Inference on LibTorch backend

We provide a notebook to demonstrate how the model is transformed into torchscript. And we provide an C++ example of how to infer with the transformed torchscript model. For details see the GitHub Actions.

Inference on ONNXRuntime backend

On the ONNXRuntime front you can use the C++ example, and we also provide a tutorial export-onnx-inference-onnxruntime for using the ONNXRuntime.

Inference on TensorRT backend

On the TensorRT front you can use the C++ example, and we also provide a tutorial onnx-graphsurgeon-inference-tensorrt for using the TensorRT.

🎨 Model Graph Visualization

Now, yolort can draw the model graph directly, checkout our model-graph-visualization notebook to see how to use and visualize the model graph.

🎓 Acknowledgement

• The implementation of yolov5 borrow the code from ultralytics.
• This repo borrows the architecture design and part of the code from torchvision.

📖 Citing yolort

If you use yolort in your publication, please cite it by using the following BibTeX entry.

@Misc{yolort2021,
  author =       {Zhiqiang Wang, Shiquan Yu, Fidan Kharrasov},
  title =        {yolort: A runtime stack for object detection on specialized accelerators},
  howpublished = {\url{https://github.com/zhiqwang/yolov5-rt-stack}},
  year =         {2021}
}

👋 Contributing

See the CONTRIBUTING file for how to help out. BTW, leave a 🌟 if you liked it, and this is the easiest way to support us :)