Implementation:Roboflow Rf detr Pip Install ONNX Export

Knowledge Sources	RF-DETR RF-DETR Export
Domains	Deployment, Model_Export
Last Updated	2026-02-08 15:00 GMT

Overview

External tool documentation for installing ONNX export dependencies for the RF-DETR library.

Description

The rfdetr[onnxexport] optional dependency group includes all packages needed for ONNX export, simplification, and validation. This includes onnx, onnxsim, onnxruntime (CPU and GPU), and graph optimization tools.

Usage

Run the pip install command once before attempting any ONNX export operation.

Code Reference

Source Location

• Repository: rf-detr
• File: rfdetr/deploy/requirements.txt
• Lines: L1-8

Command

pip install rfdetr[onnxexport]

Required Packages

Package	Purpose
onnx	Core ONNX format and model validation
onnxsim	ONNX model simplification
onnxruntime	CPU inference runtime
onnxruntime-gpu	GPU inference runtime
onnx_graphsurgeon	Graph manipulation for optimization
tensorrt>=8.6.1	NVIDIA TensorRT engine (optional)
polygraphy	TensorRT integration utilities
pycuda	CUDA Python bindings for TensorRT

I/O Contract

Inputs

Name	Type	Required	Description
Python environment	Environment	Yes	Python environment with base rfdetr package installed

Outputs

Name	Type	Description
Installed packages	Environment	ONNX export dependencies available in the Python environment

Usage Examples

Install Export Dependencies

# Install from PyPI
pip install rfdetr[onnxexport]

# Verify installation
python -c "import onnx; import onnxsim; import onnxruntime; print('ONNX export ready')"

Related Pages

Implements Principle

• Principle:Roboflow_Rf_detr_ONNX_Export_Dependencies

Requires Environment

• Environment:Roboflow_Rf_detr_ONNX_Export_Environment