Skip to content

export.md

Latest commit

 

History

History
138 lines (100 loc) · 3.86 KB

export.md

File metadata and controls

138 lines (100 loc) · 3.86 KB

RF-DETR Export Instructions

Follow the procedure listed at https://rfdetr.roboflow.com/learn/deploy/

Requirements

Important

  • Python version: 3.11 or lower (onnxsim currently requires Python <= 3.11)
  • Starting with RF-DETR 1.2.0, you must run pip install rfdetr[onnxexport] before exporting
  • Tested version: rfdetr[onnxexport]==1.3.0

Setup Virtual Environment

# install python3.11 on Ubuntu 24.04
sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt update
sudo apt install python3.11

sudo apt install python3.11-venv python3.11-distutils -y

# Create virtual environment with Python 3.11
python3.11 -m venv rfdetr_venv
source rfdetr_venv/bin/activate

# Install RF-DETR with export dependencies (tested version)
pip install rfdetr[onnxexport]==1.3.0

Detection Model Export

ONNX Export for ONNX Runtime

RF-DETR supports exporting detection models to the ONNX format, which enables interoperability with various inference frameworks and can improve deployment efficiency.

from rfdetr import RFDETRBase # or RFDETRBase/Nano/Small/Medium/Large

model = RFDETRBase(pretrain_weights=<CHECKPOINT_PATH>)

model.export()

Model Outputs:

  • dets: Bounding boxes [batch, num_queries, 4] in cxcywh format (normalized)
  • labels: Class logits [batch, num_queries, num_classes]

This command saves the ONNX model to the output directory.

Segmentation Model Export

ONNX Export for Instance Segmentation

For instance segmentation, use the RFDETRSegPreview model class or the provided export script.

Using Python Script

python deploy/export_segmentation.py --simplify --input_size 432

Available Options:

  • --output_dir: Path to save exported model (default: current directory)
  • --opset_version: ONNX opset version (default: 17)
  • --simplify: Simplify ONNX model using onnxsim
  • --batch_size: Batch size for export (default: 1)
  • --input_size: Input image size (default: 640)

Using Python API

from rfdetr import RFDETRSegPreview

model = RFDETRSegPreview(pretrain_weights=<CHECKPOINT_PATH>)

model.export(
    opset_version=17,
    simplify=True,
    batch_size=1
)

Model Outputs:

  • dets: Bounding boxes [batch, num_queries, 4] in cxcywh format (normalized)
  • labels: Class logits [batch, num_queries, num_classes]
  • masks: Segmentation masks [batch, num_queries, mask_h, mask_w] (e.g., 108x108)

This command saves the ONNX segmentation model to the output directory.

TensorRT Export (Optional)

For GPU deployment, you can convert the ONNX model to TensorRT format for optimized performance.

Detection or Segmentation Models

trtexec --onnx=/path/to/model.onnx \
        --saveEngine=/path/to/model.engine \
        --memPoolSize=workspace:4096 \
        --fp16 \
        --useCudaGraph \
        --useSpinWait \
        --warmUp=500 \
        --avgRuns=1000 \
        --duration=10

Using TensorRT Docker Container

export NGC_TAG_VERSION=25.09

docker run --rm -it --gpus=all \
    -v $(pwd)/exports:/exports \
    --ipc=host \
    --ulimit memlock=-1 \
    --ulimit stack=67108864 \
    -v $(pwd)/model.onnx:/workspace/model.onnx \
    -w /workspace \
    nvcr.io/nvidia/tensorrt:${NGC_TAG_VERSION}-py3 \
    /bin/bash -cx "trtexec --onnx=model.onnx \
                            --saveEngine=/exports/model.engine \
                            --memPoolSize=workspace:4096 \
                            --fp16 \
                            --useCudaGraph \
                            --useSpinWait \
                            --warmUp=500 \
                            --avgRuns=1000 \
                            --duration=10"

Note

TensorRT optimization works for both detection and segmentation models. The C++ inference engine supports both ONNX Runtime and TensorRT backends with compile-time backend selection.