ONNX-PREDICT-YOLOV8

This repository is a light weight library to ease the use of ONNX models exported by the Ultralytics YOLOv8 framework.

Example Usage

from onnxruntime import InferenceSession
from PIL import Image
from opyv8 import Predictor

model = Path("path/to/file.onnx")
# List of classes where the index match the class id in the ONNX network
classes = model.parent.joinpath("classes.names").read_text().split("\n")
session = InferenceSession(
    model.as_posix(),
    providers=[
        "CUDAExecutionProvider",
        "CPUExecutionProvider",
    ],
)
predictor = Predictor(session, classes)
img = Image.open("path/to/image.jpg")
print(predictor.predict(img))