🐞 <title>unspecified behavior of output names when export to Onnx

[Matthew-DgNg]

Describe the bug

I have a differences between convert to onnx when using same methods but 1 case in source code, 1 case run in .exe. Case run in .exe have the different output names, making it difficult for me to control output. I expected it will have "anomaly_map" but "onnx::Mul.."

Dataset

N/A

Model

N/A

Steps to reproduce the behavior

def anomaly2onnx(engine: Engine, model: AnomalibModule, imgWidth: int, imgHeight:int, out_dir: str = None):
path_onnx = out_dir
engine.export(
model=model.to("cuda"),
export_root=f"{path_onnx}\f32",
input_size=(imgHeight, imgWidth), # Adjust based on your needs
export_type=ExportType.ONNX
)

OS information

OS information:

• Python version: [3.10.0]
• Anomalib version: [1.1.1]
• PyTorch version: [2.8.0+cu128]
• GPU models and configuration: [e.g. 2x GeForce RTX 5060Ti]
• Any other relevant information: [e.g. I'm using a custom dataset]

Expected behavior

I have a differences between convert to onnx when using same methods but 1 case in source code, 1 case run in .exe. Case run in .exe have the different output names, making it difficult for me to control output. I expected it will have "anomaly_map" but "onnx::Mul.."

Screenshots

No response

Pip/GitHub

GitHub

What version/branch did you use?

main

Configuration YAML

None

Logs

None

Code of Conduct

• I agree to follow this project's Code of Conduct

[github-actions]

Thanks for submitting this issue! It has been added to our triage queue. A maintainer will review it shortly.