Skip to content

Model Export with Ultralytics YOLO

Ultralytics YOLO ecosystem and integrations

Introduction

The ultimate goal of training a model is to deploy it for real-world applications. Export mode in Ultralytics YOLO11 offers a versatile range of options for exporting your trained model to different formats, making it deployable across various platforms and devices. This comprehensive guide aims to walk you through the nuances of model exporting, showcasing how to achieve maximum compatibility and performance.



Watch: How To Export Custom Trained Ultralytics YOLO Model and Run Live Inference on Webcam.

Why Choose YOLO11's Export Mode?

  • Versatility: Export to multiple formats including ONNX, TensorRT, CoreML, and more.
  • Performance: Gain up to 5x GPU speedup with TensorRT and 3x CPU speedup with ONNX or OpenVINO.
  • Compatibility: Make your model universally deployable across numerous hardware and software environments.
  • Ease of Use: Simple CLI and Python API for quick and straightforward model exporting.

Key Features of Export Mode

Here are some of the standout functionalities:

  • One-Click Export: Simple commands for exporting to different formats.
  • Batch Export: Export batched-inference capable models.
  • Optimized Inference: Exported models are optimized for quicker inference times.
  • Tutorial Videos: In-depth guides and tutorials for a smooth exporting experience.

Tip

Usage Examples

Export a YOLO11n model to a different format like ONNX or TensorRT. See the Arguments section below for a full list of export arguments.

Example

from ultralytics import YOLO

# Load a model
model = YOLO("yolo11n.pt")  # load an official model
model = YOLO("path/to/best.pt")  # load a custom trained model

# Export the model
model.export(format="onnx")

yolo export model=yolo11n.pt format=onnx  # export official model
yolo export model=path/to/best.pt format=onnx  # export custom trained model

Arguments

This table details the configurations and options available for exporting YOLO models to different formats. These settings are critical for optimizing the exported model's performance, size, and compatibility across various platforms and environments. Proper configuration ensures that the model is ready for deployment in the intended application with optimal efficiency.

Argument Type Default Description
format str 'torchscript' Target format for the exported model, such as 'onnx', 'torchscript', 'engine' (TensorRT), or others. Each format enables compatibility with different deployment environments.
imgsz int or tuple 640 Desired image size for the model input. Can be an integer for square images (e.g., 640 for 640×640) or a tuple (height, width) for specific dimensions.
keras bool False Enables export to Keras format for TensorFlow SavedModel, providing compatibility with TensorFlow serving and APIs.
optimize bool False Applies optimization for mobile devices when exporting to TorchScript, potentially reducing model size and improving inference performance. Not compatible with NCNN format or CUDA devices.
half bool False Enables FP16 (half-precision) quantization, reducing model size and potentially speeding up inference on supported hardware. Not compatible with INT8 quantization or CPU-only exports for ONNX.
int8 bool False Activates INT8 quantization, further compressing the model and speeding up inference with minimal accuracy loss, primarily for edge devices. When used with TensorRT, performs post-training quantization (PTQ).
dynamic bool False Allows dynamic input sizes for ONNX, TensorRT and OpenVINO exports, enhancing flexibility in handling varying image dimensions. Automatically set to True when using TensorRT with INT8.
simplify bool True Simplifies the model graph for ONNX exports with onnxslim, potentially improving performance and compatibility with inference engines.
opset int None Specifies the ONNX opset version for compatibility with different ONNX parsers and runtimes. If not set, uses the latest supported version.
workspace float or None None Sets the maximum workspace size in GiB for TensorRT optimizations, balancing memory usage and performance. Use None for auto-allocation by TensorRT up to device maximum.
nms bool False Adds Non-Maximum Suppression (NMS) to the exported model when supported (see Export Formats), improving detection post-processing efficiency. Not available for end2end models.
batch int 1 Specifies export model batch inference size or the maximum number of images the exported model will process concurrently in predict mode. For Edge TPU exports, this is automatically set to 1.
device str None Specifies the device for exporting: GPU (device=0), CPU (device=cpu), MPS for Apple silicon (device=mps) or DLA for NVIDIA Jetson (device=dla:0 or device=dla:1). TensorRT exports automatically use GPU.
data str 'coco8.yaml' Path to the dataset configuration file (default: coco8.yaml), essential for INT8 quantization calibration. If not specified with INT8 enabled, a default dataset will be assigned.

Adjusting these parameters allows for customization of the export process to fit specific requirements, such as deployment environment, hardware constraints, and performance targets. Selecting the appropriate format and settings is essential for achieving the best balance between model size, speed, and accuracy.

Export Formats

Available YOLO11 export formats are in the table below. You can export to any format using the format argument, i.e. format='onnx' or format='engine'. You can predict or validate directly on exported models, i.e. yolo predict model=yolo11n.onnx. Usage examples are shown for your model after export completes.

Format format Argument Model Metadata Arguments
PyTorch - yolo11n.pt ✅ -
TorchScript torchscript yolo11n.torchscript ✅ imgsz, optimize, nms, batch
ONNX onnx yolo11n.onnx ✅ imgsz, half, dynamic, simplify, opset, nms, batch
OpenVINO openvino yolo11n_openvino_model/ ✅ imgsz, half, dynamic, int8, nms, batch, data
TensorRT engine yolo11n.engine ✅ imgsz, half, dynamic, simplify, workspace, int8, nms, batch, data
CoreML coreml yolo11n.mlpackage ✅ imgsz, half, int8, nms, batch
TF SavedModel saved_model yolo11n_saved_model/ ✅ imgsz, keras, int8, nms, batch
TF GraphDef pb yolo11n.pb ❌ imgsz, batch
TF Lite tflite yolo11n.tflite ✅ imgsz, half, int8, nms, batch, data
TF Edge TPU edgetpu yolo11n_edgetpu.tflite ✅ imgsz
TF.js tfjs yolo11n_web_model/ ✅ imgsz, half, int8, nms, batch
PaddlePaddle paddle yolo11n_paddle_model/ ✅ imgsz, batch
MNN mnn yolo11n.mnn ✅ imgsz, batch, int8, half
NCNN ncnn yolo11n_ncnn_model/ ✅ imgsz, half, batch
IMX500 imx yolov8n_imx_model/ ✅ imgsz, int8, data
RKNN rknn yolo11n_rknn_model/ ✅ imgsz, batch, name

FAQ

How do I export a YOLO11 model to ONNX format?

Exporting a YOLO11 model to ONNX format is straightforward with Ultralytics. It provides both Python and CLI methods for exporting models.

Example

from ultralytics import YOLO

# Load a model
model = YOLO("yolo11n.pt")  # load an official model
model = YOLO("path/to/best.pt")  # load a custom trained model

# Export the model
model.export(format="onnx")

yolo export model=yolo11n.pt format=onnx  # export official model
yolo export model=path/to/best.pt format=onnx  # export custom trained model

For more details on the process, including advanced options like handling different input sizes, refer to the ONNX integration guide.

What are the benefits of using TensorRT for model export?

Using TensorRT for model export offers significant performance improvements. YOLO11 models exported to TensorRT can achieve up to a 5x GPU speedup, making it ideal for real-time inference applications.

  • Versatility: Optimize models for a specific hardware setup.
  • Speed: Achieve faster inference through advanced optimizations.
  • Compatibility: Integrate smoothly with NVIDIA hardware.

To learn more about integrating TensorRT, see the TensorRT integration guide.

How do I enable INT8 quantization when exporting my YOLO11 model?

INT8 quantization is an excellent way to compress the model and speed up inference, especially on edge devices. Here's how you can enable INT8 quantization:

Example

from ultralytics import YOLO

model = YOLO("yolo11n.pt")  # Load a model
model.export(format="engine", int8=True)

yolo export model=yolo11n.pt format=engine int8=True   # export TensorRT model with INT8 quantization

INT8 quantization can be applied to various formats, such as TensorRT, OpenVINO, and CoreML. For optimal quantization results, provide a representative dataset using the data parameter.

Why is dynamic input size important when exporting models?

Dynamic input size allows the exported model to handle varying image dimensions, providing flexibility and optimizing processing efficiency for different use cases. When exporting to formats like ONNX or TensorRT, enabling dynamic input size ensures that the model can adapt to different input shapes seamlessly.

To enable this feature, use the dynamic=True flag during export:

Example

from ultralytics import YOLO

model = YOLO("yolo11n.pt")
model.export(format="onnx", dynamic=True)

yolo export model=yolo11n.pt format=onnx dynamic=True

Dynamic input sizing is particularly useful for applications where input dimensions may vary, such as video processing or when handling images from different sources.

What are the key export arguments to consider for optimizing model performance?

Understanding and configuring export arguments is crucial for optimizing model performance:

  • format: The target format for the exported model (e.g., onnx, torchscript, tensorflow).
  • imgsz: Desired image size for the model input (e.g., 640 or (height, width)).
  • half: Enables FP16 quantization, reducing model size and potentially speeding up inference.
  • optimize: Applies specific optimizations for mobile or constrained environments.
  • int8: Enables INT8 quantization, highly beneficial for edge AI deployments.

For deployment on specific hardware platforms, consider using specialized export formats like TensorRT for NVIDIA GPUs, CoreML for Apple devices, or Edge TPU for Google Coral devices.

📅 Created 1 year ago ✏️ Updated 0 days ago
glenn-jocher Y-T-G UltralyticsAssistant jk4e MatthewNoyce RizwanMunawar ambitious-octopus Burhan-Q Kayzwer

Comments