Roboflow ONNX

Secure model verification for Roboflow ONNX models

ONNX models exported from Roboflow can be verified before creating an inference session with the secure loader:

import onnxruntime as ort
from thistle_secure_loader import secure_load

def onnx_loader(path: str):
    return ort.InferenceSession(path)

session = secure_load("model.onnx", onnx_loader)
print("ONNX model verified.")

The secure_load call verifies the .onnx file’s signature using tuc before passing it to ort.InferenceSession. If the signature check fails, a ModelVerificationError is raised and the model is never loaded.

Requirements

onnxruntime>=1.15.0