Implementing YOLOv8 Object Detection with OpenCV in Rust Using ONNX Models
Ultralytics released the YOLOv8 two months ago (Jan 2023), it contains out-of-the-box support for object detection, classification, and segmentation. As YOLO is popular and well-known, further details are omitted here. YOLO is very easy to use in Python, just a few lines of code, and we can get the detection results.
ONNX is a standard format for machine learning models, which is supported by many ML frameworks. ONNX is also supported by OpenCV, so we can use OpenCV to load the ONNX model and do inference. OpenCV is orginally written in C++ which I am familiar with, but it also provides Rust bindings. In this post, I will use Rust to implement YOLOv8 object detection with OpenCV using ONNX models (no ONNXRuntime), more for my own reference.
You can find the the Python implementation of YOLOv8 on OpenCV using ONNX in the official repository .
Install Rust by following the official instructions .
If we want to use OpenCV crate in Rust , it requires OpenCV system library to be installed in the system. It’s not complicated, we have no need to build OpenCV from source like in C++. Follow the instructions to set it up in a few minutes.