Arm/pidnet-s-int8-xnnpack-executorch-raspberrypi5

PIDNet small optimized for Arm-based Edge Linux systems

A semantic segmentation model quantized to INT8 via static post-training quantization — per-channel symmetric weights, per-tensor affine activations — exported to ExecuTorch (.pte) and optimized for Arm-based Edge Linux systems.

Summary

This repository contains an Arm-optimized version of pidnet_s for semantic segmentation of urban street scenes, quantized to INT8 via static post-training quantization — per-channel symmetric weights, per-tensor affine activations. 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 Cityscapes 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	p50 latency 956.286 ms (3.48x faster than baseline), 1.05 frames per second
Accuracy result	Mean IoU 78.6%
Size / memory result	7.536 MB (3.87x smaller), peak memory 290.23 MB

Original model

Field	Value
Original model	pidnet_s
Original source	GitHub
Original developer	PIDNet (Jiacong Xu, Zixiang Xiong and Shankar P. Bhattacharyya)
Original model card	PIDNet repository
Original license	MIT

Model files

File	Description
pidnet_s_raspberry_executorch_optimized.pte	Arm-optimized INT8 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 @ 2.4 GHz
System memory	8 GB
OS	Linux, Raspberry Pi OS 64-bit, based on Debian 13 "Trixie"
Runtime	ExecuTorch 1.1.0
Backend / delegate	XNNPACK, KleidiAI
Threads	1
Batch size	1
Input resolution	2048x1024
Precision	INT8, static PTQ — per-channel symmetric weights, per-tensor affine activations
Runs	5 warmup + 20 measured

Performance results

Metric	Original / baseline	Arm-optimized	Improvement
Model size (MB)	29.173	7.536	3.87x smaller
End-to-end latency p50 (ms)	3324.178	956.286	3.48x faster
End-to-end latency p90 (ms)	3331.712	961.969	3.46x faster
End-to-end latency p99 (ms)	3336.645	965.431	3.46x faster
Frames per second	0.3	1.05	3.50x
Model load time (ms)	47.83	18.039	2.65x faster
Time to first inference (ms)	3515.942	1032.721	3.40x faster
Peak memory (MB)	545.95	290.23	1.88x less
Average memory (MB)	545.42	289.67	1.88x 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	Cityscapes
Split	val
Number of samples	500
Metric(s)	Mean IoU
Evaluation runtime	ExecuTorch 1.1.0

Accuracy results

Metric	Original / baseline	Arm-optimized	Change
Mean IoU (%)	78.76	78.6	-0.16 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
Quantization	Yes	INT8 PTQ-static — per-channel symmetric weights, per-tensor affine activations; calibrated on 64 randomly selected Cityscapes samples
Runtime/backend selection	Yes	XNNPACK + KleidiAI delegate on CPU
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, 1024, 2048]
Input type	float32
Input range	[0.0, 1.0]
Preprocessing	preprocess() in example.py

Expected output

Property	Value
Output shape	[1, 19, 128, 256]
Output type	Per-pixel class logits at 1/8 input resolution, 19 Cityscapes trainId classes
Postprocessing	postprocess() in example.py

Intended use

This model is intended for developers evaluating semantic segmentation 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 the Cityscapes val split 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

Quantization was performed with PT2E static quantization — INT8, per-channel symmetric weights, per-tensor affine activations. All layers were quantized; no layers were kept in FP32, and no graph surgery was required. Calibration used randomly selected images from the Cityscapes train split, disjoint from the val split used for evaluation (the two splits share no cities).

The evaluation protocol upsamples the logits before the argmax — rather than resizing the class map afterwards — and uses align_corners true. Both are part of the published single-scale Mean IoU protocol; changing either shifts the sampling grid and costs accuracy.

The model uses a fixed 2048x1024 input with a simple resize — no letterboxing or cropping — so images of other aspect ratios are stretched. It predicts only the 19 Cityscapes trainId classes and was trained on European urban street scenes, so anything outside that label set or domain is forced into one of those classes.

• Sample input: sample_input.jpg is derived from Rue Bouvier, Quebec City (17) by Wilfredor, via Wikimedia Commons (CC0 1.0).

About this version

Original Model: pidnet_s by PIDNet (Jiacong Xu, Zixiang Xiong and Shankar P. Bhattacharyya) - 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 MIT.

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.