Arm/yolo11n-pose-int8-xnnpack-executorch-raspberrypi5

YOLO11n-Pose optimized for Arm-based Edge Linux

YOLO11n-Pose optimized for pose estimation, exported to ExecuTorch (.pte format), targeting Edge Linux systems.

Summary

This repository contains an Arm-optimized version of yolo11n-pose for pose estimation (keypoint 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 (pose) 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, Raspberry Pi OS 64-bit based on Debian 13 "Trixie")
Primary performance result	234.00 ms p50 latency (4.27 FPS)
Accuracy result	46.38% mAP@0.5:0.95
Size / memory result	5.13 MB (2.18x smaller than the 11.17 MB baseline)

Original model

Field	Value
Original model	yolo11n-pose
Original source	GitHub
Original developer	Ultralytics
Original model card	Ultralytics/YOLO11
Original license	AGPL-3.0

Model files

File	Description
yolo11n-pose_raspberry_executorch_optimized.pte	Arm-optimized model for deployment
example.py	Minimal inference example
pyproject.toml	Pinned runtime dependencies for example.py, resolved with uv
uv.lock	Locked dependency resolution for pyproject.toml
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 @ 2.4GHz)
OS	Raspberry Pi OS 64-bit based on Debian 13 "Trixie"
Runtime	ExecuTorch 1.1.0
Backend / delegate	XNNPACK, KleidiAI
Batch size	1
Precision	INT8 static PTQ — per-channel symmetric weights, per-tensor affine activations
Runs	10 warmup + 100 measured

Performance results

Metric	Original / baseline	Arm-optimized	Improvement
p50 latency	335.84 ms	234.00 ms	1.44x faster
p90 latency	336.10 ms	234.18 ms	1.44x faster
p99 latency	337.40 ms	234.35 ms	1.44x faster
Model size	11.17 MB	5.13 MB	2.18x smaller
Peak memory	75.39 MB	59.19 MB	1.27x less
Frames per second	2.98	4.27	1.43x

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 (pose)
Split	N/A
Number of samples	5000
Metric(s)	mAP@0.5:0.95, mAP@0.5, mAP@0.75
Evaluation runtime	ExecuTorch

Accuracy results

Metric	Original / baseline	Arm-optimized	Change
mAP@0.5:0.95	47.37%	46.38%	-0.99 pp
mAP@0.5	75.95%	75.79%	-0.16 pp
mAP@0.75	50.68%	49.62%	-1.06 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	Exported to ExecuTorch (.pte format) via the shared PT2E capture/prepare/convert pipeline.
Quantization	Yes	INT8 static PTQ — per-channel symmetric weights, per-tensor affine activations; calibrated on 1,000 randomly sampled COCO (pose) training images.
Runtime/backend selection	Yes	XNNPACK with KleidiAI kernels on ExecuTorch's CPU backend.
Graph/runtime compatibility updates	Yes	Performed as part of the ExecuTorch export pipeline.
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

Dependencies are declared in pyproject.toml, which ships with this repository. Resolve and install them into a local virtual environment with uv:

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, pad value 114,114,114), convert to tensor — no normalization

Expected output

Property	Value
Output shape	[1, 56, N] (transposed to [1, N, 56] before decoding; N depends on input size)
Output type	N/A
Postprocessing	Compute class confidence score (sigmoid applied only if raw values fall outside [0, 1]); convert boxes from xywh (center) to xyxy; filter detections by confidence threshold 0.25; apply batched NMS per class at IoU threshold 0.45; reshape keypoints to [N, 17, 3] as (x, y, visibility) in the 640x640 letterboxed coordinate space

Intended use

This model is intended for developers evaluating pose estimation / keypoint 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 val (pose), 5,000 images and may not generalize to all domains.
• Requires a fixed 640x640 input; images are letterboxed with gray padding to preserve aspect ratio rather than stretched.
• This repository is not a replacement for the original model documentation.

Additional notes

The pose head (model.23) is kept out of INT8 quantization: its output concatenates box coordinates, a class score, and keypoint x/y/visibility triplets at very different value ranges, and a single per-tensor quantizer collapses the visibility channel to a constant. Post-convert graph surgery strips the Q/DQ pairs around the pose head's cat and reshape nodes so the final output stays in floating point. The first convolution layer is also protected and kept at higher precision. Calibration used 1,000 randomly sampled images from the COCO (pose) training split.

• Sample input: sample_input.jpg is derived from Walking Rugby by Back ache, via Wikimedia Commons (CC0 1.0).

About this version

Original Model: yolo11n-pose 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.

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.