search

FourierMourier / yolov8-onnx-cpp

YOLOv8 C++ Instance Segmentation, Pose Estimation and Object Detection with ONNX
Apache License 2.0
103 stars 23 forks source link

readme

# yolov8-onnx-cpp

Changelog

See the changelog here

Description

Hello there! yolov8-onnx-cpp is a C++ demo implementation of the YOLOv8 model using the ONNX library. This project is based on the YOLOv8 model by Ultralytics. I aimed to replicate the behavior of the Python version and achieve consistent results across various image sizes.

By the way, you don't need to specify names, img_size etc while initializing the model, since we can use ONNX metadata!

When you do export in python in onnx format, the following code executes

    self.metadata = {
        'description': description,
        'author': 'Ultralytics',
        'license': 'AGPL-3.0 https://ultralytics.com/license',
        'date': datetime.now().isoformat(),
        'version': __version__,
        'stride': int(max(model.stride)),
        'task': model.task,
        'batch': self.args.batch,
        'imgsz': self.imgsz,
        'names': model.names}  # model metadata
    if model.task == 'pose':
        self.metadata['kpt_shape'] = model.model[-1].kpt_shape

(ultralytics 8.0.160, ultralytics/engine/exporter.py lines 221-233))

We can use this parameters at least to define stride, task, names and image size as described in the schema below:

Schema

Supported Tasks and Hardware

Task Supported
Detect ✔️
Segment ✔️
Pose ✔️
Classify
Hardware Supported
CPU ✔️
GPU

Comparison between Python and C++

I exported yolov8n.pt, yolov8n-seg.pt, yolov8n-pose.pt to ONNX format with an input size of [480, 640] ([height, width]). For the test I used some images from the COCO128 dataset with different image sizes than the specified input. This difference in sizes triggered letterboxing. I maintained consistent parameters, setting conf=0.3 and iou=0.5 for all models.

Here are the comparison results:

Pose

Python Result

Python Result

C++ Result

C++ Result

Segmentation

Python Result 1

Python Result

C++ Result 1

C++ Result

Python Result 2

Python Result

C++ Result 2

C++ Result

Object detection

Python Result 1

Python Result

C++ Result 1

C++ Result

Python Result 2

Python Result

C++ Result 2

C++ Result

Getting Started

To get started with yolov8-onnx-cpp, follow these steps:

  1. Clone the repository: 
    git clone https://github.com/FourierMourier/yolov8-onnx-cpp.git
  2. Setup additional libraries:

Download opencv here (4.80+)

Solution-like build

Cmake in non-visual studio-like ide

Copy CMakeLists.example.txt as CMakeLists.txt (git-ignored) and edit the following lines:


SET (OpenCV_DIR your/path/to/opencv/build/x64/vc16/lib)  # opencv lib root
SET (OpenCV_BIN_DIR your/path/to/opencv/build/x64/vc16/bin)  #opencv bin root

SET (OpenCV_DEBUG_DLL_FILENAME opencv_world480d.dll)  # change filenames
SET (OpenCV_RELEASE_DLL_FILENAME opencv_world480.dll)  # change filenames

SET (ONNXRUNTIME_DIR your/path/to/onnxruntime-win-x64-1.15.1)  # onnxruntime root

Even though you'll find only a nuget package on the official page you can still download release for cmake file here for your platform: https://github.com/microsoft/onnxruntime/releases

If you're working in different IDE like Clion rather than visual studio you still have to do the following:

So the issues like

The issue like "The given version [15] is not supported, only version 1 to 10 is supported in this build=" https://github.com/microsoft/onnxruntime/issues/11230

also should not occur since you configure dll in cmake

Hope that helps!

  1. edit in the ./src/main.cpp img_path/&modelPath: 
    std::string img_path = "./images/000000000143.jpg";
//const std::string& modelPath = "./checkpoints/yolov8n.onnx"; // detection
const std::string& modelPath = "./checkpoints/yolov8n-seg.onnx"; // instance segmentation
const std::string& onnx_provider = OnnxProviders::CPU; // "cpu";

    Usage

    Provide an input image to the application, and it will perform object detection using the YOLOv8 model. Customize the model configuration and parameters in the code as needed.

References

License

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.

Acknowledgments

Some other cool repositories I found useful (and you might too):

This README was created with the assistance of OpenAI's ChatGPT (August 3 Version), a large language model. You can learn more about it here