YOLO to ONNX Conversion - Revision history

Giuseppe at 07:44, 26 June 2026

2026-06-26T07:44:25Z

← Older revision		Revision as of 07:44, 26 June 2026
Line 7:		Line 7:
	If you require professional assistance, please contact <b>sales@logicaldoc.com</b> to request a quotation for consulting services.}}		If you require professional assistance, please contact <b>sales@logicaldoc.com</b> to request a quotation for consulting services.}}

	{{~~Advice~~\| <b>Licensing ~~notice~~ </b>: Ultralytics YOLO is distributed under its own licensing terms. <b><u>~~Commercial use requires~~ a commercial license ~~from Ultralytics~~.</u></b> ~~Before using YOLO in a commercial environment, review~~ the licensing information ~~available on~~ the official Ultralytics website: https://www.ultralytics.com/}}		{{Warning\|
			<b>Licensing Notice:</b> The examples in this guide use the Ultralytics implementation of YOLO. Ultralytics YOLO is distributed under its own licensing terms. <b><u>Before using YOLO in a commercial environment, ensure that your intended use complies with the applicable Ultralytics license.</u></b> For the latest licensing information, refer to the official Ultralytics website: https://www.ultralytics.com/license
			}}

	LogicalDOC performs object detection using the ONNX Runtime Java API. Therefore, a trained YOLO model must be converted from the native PyTorch (`.pt`) format to the Open Neural Network Exchange (ONNX) format before it can be imported into LogicalDOC.		LogicalDOC performs object detection using the ONNX Runtime Java API. Therefore, a trained YOLO model must be converted from the native PyTorch (`.pt`) format to the Open Neural Network Exchange (ONNX) format before it can be imported into LogicalDOC.

Giuseppe at 07:39, 26 June 2026

2026-06-26T07:39:15Z

← Older revision		Revision as of 07:39, 26 June 2026
Line 3:		Line 3:
	== Converting a YOLO Model to ONNX ==		== Converting a YOLO Model to ONNX ==

	{{Advice\|		{{Advice\| This guide describes an example workflow for training a custom YOLO model and preparing it for use with LogicalDOC.
	<b><u>Please be aware that this procedure is not coverded by the standard support contract</u></b>. LogicalDOC cannot provide assistance with issues related to dataset preparation, training failures, model quality, GPU configuration, or third-party tools such as Label Studio, Ultralytics YOLO, or ONNX Runtime.		<b><u>Please be aware that this procedure is not coverded by the standard support contract</u></b>. LogicalDOC cannot provide assistance with issues related to dataset preparation, training failures, model quality, GPU configuration, or third-party tools such as Label Studio, Ultralytics YOLO, or ONNX Runtime.
	If you require professional assistance, please contact sales@logicaldoc.com to request a quotation for consulting services.}}		If you require professional assistance, please contact <b>sales@logicaldoc.com</b> to request a quotation for consulting services.}}

			{{Advice\| <b>Licensing notice </b>: Ultralytics YOLO is distributed under its own licensing terms. <b><u>Commercial use requires a commercial license from Ultralytics.</u></b> Before using YOLO in a commercial environment, review the licensing information available on the official Ultralytics website: https://www.ultralytics.com/}}

	LogicalDOC performs object detection using the ONNX Runtime Java API. Therefore, a trained YOLO model must be converted from the native PyTorch (`.pt`) format to the Open Neural Network Exchange (ONNX) format before it can be imported into LogicalDOC.		LogicalDOC performs object detection using the ONNX Runtime Java API. Therefore, a trained YOLO model must be converted from the native PyTorch (`.pt`) format to the Open Neural Network Exchange (ONNX) format before it can be imported into LogicalDOC.

Giuseppe at 07:33, 26 June 2026

2026-06-26T07:33:50Z

← Older revision		Revision as of 07:33, 26 June 2026
Line 2:		Line 2:

	== Converting a YOLO Model to ONNX ==		== Converting a YOLO Model to ONNX ==

			{{Advice\|
			<b><u>Please be aware that this procedure is not coverded by the standard support contract</u></b>. LogicalDOC cannot provide assistance with issues related to dataset preparation, training failures, model quality, GPU configuration, or third-party tools such as Label Studio, Ultralytics YOLO, or ONNX Runtime.
			If you require professional assistance, please contact sales@logicaldoc.com to request a quotation for consulting services.}}


	LogicalDOC performs object detection using the ONNX Runtime Java API. Therefore, a trained YOLO model must be converted from the native PyTorch (`.pt`) format to the Open Neural Network Exchange (ONNX) format before it can be imported into LogicalDOC.		LogicalDOC performs object detection using the ONNX Runtime Java API. Therefore, a trained YOLO model must be converted from the native PyTorch (`.pt`) format to the Open Neural Network Exchange (ONNX) format before it can be imported into LogicalDOC.

Giuseppe: /* Export the Model */

2026-06-25T09:23:18Z

Export the Model

← Older revision		Revision as of 09:23, 25 June 2026
Line 23:		Line 23:
	== Export the Model ==		== Export the Model ==

	The following script converts the trained `best.pt` model into the ONNX format:		The following script converts the trained <code>best.pt</code> model into the ONNX format:

	<syntaxhighlight lang="python">		<syntaxhighlight lang="python">
Line 38:		Line 38:
	Where:		Where:

	* `format="onnx"` exports the model in the ONNX format.		* <code>format="onnx"</code> exports the model in the ONNX format.
	* `simplify=True` simplifies the exported computation graph to improve compatibility and potentially reduce inference overhead.		* <code>simplify=True</code> simplifies the exported computation graph to improve compatibility and potentially reduce inference overhead.
	* `nms=True` embeds the Non-Maximum Suppression (NMS) step into the exported model, allowing the ONNX Runtime to return the final detections directly.		* <code>nms=True</code> embeds the Non-Maximum Suppression (NMS) step into the exported model, allowing the ONNX Runtime to return the final detections directly.

	After the conversion completes, the generated model (`best.onnx`) can be imported into LogicalDOC for inference.		After the conversion completes, the generated model (<code>best.onnx</code> ) can be imported into LogicalDOC for inference.

Giuseppe at 09:22, 25 June 2026

2026-06-25T09:22:13Z

← Older revision		Revision as of 09:22, 25 June 2026
Line 1:		Line 1:
			= YOLO to ONNX Conversion =

	== Converting a YOLO Model to ONNX ==		== Converting a YOLO Model to ONNX ==

Giuseppe at 09:21, 25 June 2026

2026-06-25T09:21:33Z

← Older revision		Revision as of 09:21, 25 June 2026
Line 1:		Line 1:
	~~= YOLO to ONNX Conversion =~~

	== Converting a YOLO Model to ONNX ==		== Converting a YOLO Model to ONNX ==

Giuseppe: /* Converting a YOLO Model to ONNX */

2026-06-25T09:20:54Z

Converting a YOLO Model to ONNX

← Older revision		Revision as of 09:20, 25 June 2026
Line 1:		Line 1:
	= ~~Converting a~~ YOLO ~~Model~~ to ONNX =		= YOLO to ONNX Conversion =

	~~LogicalDOC performs object detection using the ONNX Runtime Java API. For this reason, trained~~ YOLO ~~models must be converted from the native PyTorch (.pt) format~~ to ~~the Open Neural Network Exchange (~~ONNX~~) format before they can be imported into LogicalDOC.~~		== Converting a YOLO Model to ONNX ==

	ONNX (Open Neural Network Exchange) ~~is an open standard for representing machine learning models. It enables models trained in one framework, such as PyTorch, to~~ be ~~executed efficiently in different environments and programming languages~~.		LogicalDOC performs object detection using the ONNX Runtime Java API. Therefore, a trained YOLO model must be converted from the native PyTorch (`.pt`) format to the Open Neural Network Exchange (ONNX) format before it can be imported into LogicalDOC.

	The conversion process does not retrain or modify the model. Instead, it exports the learned weights and computational graph into a portable format that can be executed by ~~the~~ ONNX Runtime ~~inference engine~~.		ONNX (Open Neural Network Exchange) is an open standard for representing machine learning models. It enables models trained in one framework, such as PyTorch, to be executed efficiently in different programming languages and runtime environments.

			The conversion process does not retrain or modify the model. Instead, it exports the trained network, including its learned weights and computational graph, into a portable format that can be executed by ONNX Runtime.

	== Why ONNX? ==		== Why ONNX? ==
Line 15:		Line 17:
	* No dependency on the Python runtime during inference.		* No dependency on the Python runtime during inference.
	* Native Java support through the ONNX Runtime API.		* Native Java support through the ONNX Runtime API.
	* Faster startup and lower memory consumption ~~compared to~~ embedding a Python interpreter.		* Faster startup and lower memory consumption than embedding a Python interpreter.
	* Cross-platform model portability.		* Cross-platform model portability.
	* ~~Hardware~~ acceleration when ~~supported by the execution environment~~.		* Support for hardware acceleration when available.

			== Export the Model ==

			The following script converts the trained `best.pt` model into the ONNX format:

	~~This is an example of python script that performs the YOLO-to-ONNX conversion:~~		<syntaxhighlight lang="python">

	<syntaxhighlight lang="~~Python~~">
	from ultralytics import YOLO		from ultralytics import YOLO

Line 29:		Line 31:

	model.export(		model.export(
	format="onnx",		format="onnx",
	simplify=True,		simplify=True,
	nms=True		nms=True
	)		) </syntaxhighlight>
	</syntaxhighlight>
			Where:

			* `format="onnx"` exports the model in the ONNX format.
			* `simplify=True` simplifies the exported computation graph to improve compatibility and potentially reduce inference overhead.
			* `nms=True` embeds the Non-Maximum Suppression (NMS) step into the exported model, allowing the ONNX Runtime to return the final detections directly.

			After the conversion completes, the generated model (`best.onnx`) can be imported into LogicalDOC for inference.

Giuseppe: Created page with "= Converting a YOLO Model to ONNX = LogicalDOC performs object detection using the ONNX Runtime Java API. For this reason, trained YOLO models must be converted from the native PyTorch (.pt) format to the Open Neural Network Exchange (ONNX) format before they can be imported into LogicalDOC. ONNX (Open Neural Network Exchange) is an open standard for representing machine learning models. It enables models trained in one framework, such as PyTorch, to be executed effici..."

2026-06-25T08:06:20Z

Created page with "= Converting a YOLO Model to ONNX = LogicalDOC performs object detection using the ONNX Runtime Java API. For this reason, trained YOLO models must be converted from the native PyTorch (.pt) format to the Open Neural Network Exchange (ONNX) format before they can be imported into LogicalDOC. ONNX (Open Neural Network Exchange) is an open standard for representing machine learning models. It enables models trained in one framework, such as PyTorch, to be executed effici..."

New page

= Converting a YOLO Model to ONNX =

LogicalDOC performs object detection using the ONNX Runtime Java API. For this reason, trained YOLO models must be converted from the native PyTorch (.pt) format to the Open Neural Network Exchange (ONNX) format before they can be imported into LogicalDOC.

ONNX (Open Neural Network Exchange) is an open standard for representing machine learning models. It enables models trained in one framework, such as PyTorch, to be executed efficiently in different environments and programming languages.

The conversion process does not retrain or modify the model. Instead, it exports the learned weights and computational graph into a portable format that can be executed by the ONNX Runtime inference engine.

== Why ONNX? ==

Ultralytics trains YOLO models using PyTorch. However, LogicalDOC is implemented in Java and performs inference using the ONNX Runtime library.

Using ONNX provides several advantages:

* No dependency on the Python runtime during inference.
* Native Java support through the ONNX Runtime API.
* Faster startup and lower memory consumption compared to embedding a Python interpreter.
* Cross-platform model portability.
* Hardware acceleration when supported by the execution environment.

This is an example of python script that performs the YOLO-to-ONNX conversion:

<syntaxhighlight lang="Python">
from ultralytics import YOLO

model = YOLO("runs/detect/target/weights/best.pt")

model.export(
format="onnx",
simplify=True,
nms=True
)
</syntaxhighlight>