Arm/yolo11n-int8-xnnpack-executorch-raspberrypi5

YOLO11n optimized for Arm-based Edge Linux

YOLO11n object detection optimized using static post-training quantization (PTQ), with INT8 quantization across eligible backbone and neck operations while the stem convolution (model.0.conv) and complete detection head (model.23) remain in FP32 to preserve accuracy within acceptable ranges. The model is exported to ExecuTorch .pte format for efficient inference on Arm-based Edge Linux systems using the XNNPACK backend.

Summary

This repository contains an Arm-optimized version of yolo11n for object detection. The model is provided in ExecuTorch .pte format, targeting Edge Linux systems.

This version is intended to demonstrate efficient inference on Arm-based platforms while preserving the original model's intended behavior. Arm has evaluated this model on COCO 2017 (val2017) and measured performance on a representative evaluation target.

Key results

Area	Result
Model format	ExecuTorch .pte
Target device class	Edge Linux
Reference device	Raspberry Pi 5 (Cortex-A76, Linux, Raspberry Pi OS 64-bit, based on Debian 13 "Trixie")
Primary performance result	7.46 FPS (134.10 ms p50 latency)
Accuracy result	mAP@0.5:0.95 38.90%
Size / memory result	4.73 MB, 2.16x smaller than the FP32 baseline

Original model

Field	Value
Original model	yolo11n
Original source	GitHub
Original developer	Ultralytics
Original model card	https://huggingface.co/Ultralytics/YOLO11
Original license	AGPL-3.0 (a commercial Enterprise license is also available from Ultralytics for use cases that cannot comply with AGPL-3.0)

Model files

File	Description
yolo11n_raspberry_executorch_optimized.pte	Arm-optimized model for deployment
example.py	Minimal inference example
pyproject.toml	Pinned runtime dependencies and deployment metadata
uv.lock	Locked dependency versions, sources, and hashes for uv sync --frozen
config.yaml	Model I/O contract used by the example
benchmarks/	FP32 baseline and Arm-optimized benchmark records

Performance

Performance was measured on the reference configuration below. Results are intended to make the optimization reproducible but do not guarantee identical performance on every Arm-based system.

Reference configuration

Field	Value
Device / platform	Raspberry Pi 5
CPU / accelerator	Cortex-A76 (4 cores @ 2.4 GHz), CPU
OS	Linux, Raspberry Pi OS 64-bit, based on Debian 13 "Trixie"
Runtime	ExecuTorch 1.1.0
Backend / delegate	XNNPACK, KleidiAI
Batch size	1
Precision	Mixed precision: INT8 for eligible backbone and neck operations; the stem convolution (model.0.conv) and complete detection head (model.23) remain in FP32
Quantization method	Static PTQ with symmetric, per-channel weight quantization
Runs	10 warmup + 100 measured

Performance results

Metric	Original / baseline	Arm-optimized	Improvement
p50 latency	169.36 ms	134.10 ms	1.26x faster
p90 latency	170.12 ms	134.68 ms	1.26x faster
p99 latency	171.62 ms	135.06 ms	1.27x faster
Frames per second	5.90	7.46	1.26x higher
Model size	10.23 MB	4.73 MB	2.16x smaller
Model load time	12.84 ms	8.82 ms	1.46x faster
Time to first inference	191.71 ms	150.83 ms	1.27x faster
Peak memory	78.33 MB	59.17 MB	1.32x less
Average memory	77.42 MB	57.98 MB	1.34x less

Accuracy

Accuracy was evaluated using the same preprocessing, input resolution, and evaluation protocol described below. Where possible, the optimized model is compared against the original model under the same evaluation conditions.

Evaluation setup

Field	Value
Dataset	COCO 2017
Split	val2017
Number of samples	5,000
Metric(s)	mAP@0.5:0.95, mAP@0.5, mAP@0.75
Evaluation runtime	ExecuTorch (Arm-optimized) / PyTorch (FP32 baseline)

Accuracy results

Metric	Original / baseline	Arm-optimized	Change
mAP@0.5:0.95	39.30%	38.90%	-0.40 pp
mAP@0.5	55.11%	54.66%	-0.45 pp
mAP@0.75	42.75%	42.38%	-0.37 pp

Accuracy was measured using the evaluation setup described above. Users should re-evaluate the model on their own data before production use.

Arm optimization approach

Arm optimized this model for efficient inference on Arm-based platforms using a hardware-aware conversion and validation flow.

For this release, Arm used:

Optimization area	Applied?	Notes
Model conversion	Yes	Converted to ExecuTorch .pte, targeting the XNNPACK backend
Quantization	Yes	Static INT8 PTQ with symmetric, per-channel weight quantization, calibrated on 2,000 COCO 2017 training images. The stem convolution and complete detection head (model.23) remain in FP32; remaining concatenations and their direct consumers are excluded from the standard quantization path
Runtime/backend selection	Yes	XNNPACK backend with KleidiAI kernels, 1 thread
Graph/runtime compatibility updates	Yes	Internal C3k2, C3k, SPPF, and C2PSA concatenations are rewritten before capture as independent per-branch convolutions with folded BatchNorm. After conversion, forced Q/DQ pairs are removed from remaining concatenations network-wide, including three detection-head concatenations and the final output merge, located through FX graph topology
Accuracy validation	Yes	Compared against the original model or published baseline
Performance validation	Yes	Measured on the reference Arm platform

The goal of this process is to improve deployment characteristics such as latency, memory use, model size, and runtime compatibility while preserving the model's intended behavior. Detailed conversion scripts, calibration configuration, or backend-specific implementation details may be provided separately where appropriate.

Using this model

Install dependencies

uv python install
uv sync --frozen

Run the example

uv run example.py

Expected input

Property	Value
Input shape	[1, 3, 640, 640]
Input type	float32
Input range	[0.0, 1.0]
Preprocessing	Letterbox-resize to 640x640 (aspect-ratio-preserving resize + pad, pad value 114/255, center-anchored); convert to tensor by resizing uint8 pixels first and then dividing by 255 to obtain [0, 1] float32 values (Ultralytics order); no mean/std normalization

Expected output

Property	Value
Output shape	[1, 84, 8400]
Output type	N/A
Postprocessing	Transpose to [1, 8400, 84] (4 xywh center coordinates + 80 class scores, no objectness); score = max(class_scores); filter below the confidence threshold (0.001 for the reported mAP, so the full precision-recall curve is retained; the bundled example uses 0.4 for readable output); convert xywh to xyxy; keep top 30,000 candidates by score; per-class batched NMS (IoU threshold 0.7); keep top 300 detections; rescale via the inverse letterbox transform back to original image coordinates

Intended use

This model is intended for developers evaluating object detection workloads on Arm-based platforms. It is suitable as a reference implementation for benchmarking, prototyping, and integration exploration.

Limitations

• Performance depends on the target device, runtime version, backend/delegate support, memory configuration, and system load.
• Accuracy was evaluated on COCO 2017 val2017 and may not generalize to all domains.
• This release preserves the original model's intended task and behavior, but users should validate it for their own application, data, and deployment environment.
• This repository is not a replacement for the original model documentation.

Additional notes

• Fixed input size of 640x640 with letterbox padding; dynamic or arbitrary input resolutions require re-export.
• Sample input: sample_input.jpg is derived from Living room (Unsplash) by Jarosław Ceborski, via Wikimedia Commons (CC0 1.0).

About this version

Original Model: yolo11n by Ultralytics - Repository

Optimization/conversion: Arm-Optimized version for execution on Arm-based platforms.

Converted/optimized by: Arm

License: The Original Model and the Optimized Model are subject to AGPL-3.0 (a commercial Enterprise license is also available from Ultralytics for use cases that cannot comply with AGPL-3.0).

This repository contains a converted or optimized version of the Original Model (the “Optimized Model”). The Original Model has been converted or optimized as described above for execution on Arm-based platforms.

No retraining or fine-tuning of the Original Model was performed as part of the conversion or optimization. The conversion or optimization was not intended to change the Original Model’s behavior or intended use.

Original Model and Documentation

For information about the Original Model, including its development, training data, intended uses, limitations and other relevant information, please refer to the Original Model repository. Information in that repository was provided by the original developer or other third parties and, unless expressly stated otherwise, has not been independently verified by Arm.

Licenses and Third-Party Terms

Use of the Original Model and the Optimized Model is subject to the applicable licenses, usage restrictions and other terms identified above and in the relevant repositories. Publication of the Optimized Model does not grant any rights beyond those provided under the applicable license terms.

You are responsible for reviewing those terms and ensuring that your use of the Original Model and the Optimized Model is permitted.

Purpose of this Release

The Optimized Model is provided as a reference implementation to demonstrate and evaluate execution and performance on Arm-based systems. It is not a production-ready or supported solution.

Arm’s publication of the Optimized Model does not constitute an endorsement or certification of the Original Model or a representation that the Optimized Model is suitable for production use or any particular purpose.

To the fullest extent permitted by applicable law (i) the Optimized Model is provided “as is.” Arm makes no representations or warranties that the Original Model, the Optimized Model or their outputs are accurate, safe, secure, non-infringing, legally compliant, suitable for production use or fit for any particular purpose; and (ii) Arm will not be liable for any loss or damage arising from or in connection with the Optimized Model, its use or its outputs.

You are responsible for independently evaluating the Optimized Model, its outputs and its suitability for your intended use, including compliance with applicable legal, regulatory, safety and security requirements.

Arm does not commit to provide ongoing support, maintenance or updates for the Optimized Model. Any use of or reliance on the Optimized Model or its outputs is at your own risk.

Use of Ultralytics models

Ultralytics provides Ultralytics YOLO software and models available through the Arm AI Portal and/or Arm's Hugging Face Organization under the GNU Affero General Public License v3.0 ("AGPL-3.0"), unless you have entered into a separate written license agreement with Ultralytics. Accessing, downloading, or retraining these materials through Arm AI Portal does not grant you an Ultralytics Enterprise License or any other proprietary Ultralytics license.

AGPL-3.0 requires you to release the complete source code of any application that uses Ultralytics YOLO, including applications made available over a network, under the same license. If you are embedding YOLO in a commercial product, internal tool, or production deployment and cannot open-source your code, you need an Ultralytics Enterprise License.

You are responsible for determining which applies to your use. Terms and enterprise licensing options are available at ultralytics.com.