Accelerate Denoising, Background Blur and Low-Light Camera Effects with SME2
Introduction
Prerequisites
Overview
Build the pipelines
Run the pipelines
Performance
Next Steps

  Log an issue

  Fork and edit

  Discuss on Discord

Accelerate Denoising, Background Blur and Low-Light Camera Effects with SME2

About this Learning Path

Skill level:  Introductory
Reading time:  30 min
Last updated:  05 Aug 2026
Skill level:
  Introductory
Reading time:
  30 min
Last updated:
  05 Aug 2026
Author:Arnaud de Grandmaison, Arm
Arm IP:Cortex-A Arm C1
Tags: Performance and Architecture Linux macOS CPP Docker SME2
Author:
  • Arnaud de Grandmaison, Arm
Arm IP:
Cortex-A Arm C1
Tags:
Performance and Architecture Linux macOS CPP Docker SME2

Who is this for?

This introductory topic is for mobile and computer-vision developers, camera pipeline engineers, and performance-minded practitioners who want to optimize real-time camera effects on Arm using KleidiAI and KleidiCV.

What will you learn?

Upon completion of this Learning Path, you will be able to:

  • Build and run AI-powered camera pipeline applications
  • Use SME2 to improve the performance of real-time camera pipelines

Prerequisites

Before starting, you will need the following:

  • A computer running Arm Linux or macOS with Docker installed

Summary

AI-assisted

This summary was drafted with an approved AI-assisted workflow and reviewed by Arm contributors before publication. Human technical review remains part of the process so the final page reflects engineering rigor, accuracy, and Arm editorial standards.

Close
?
You’ll clone the AI camera pipelines repository, fetch assets with Git Large File Storage (LFS), and build the projects in Docker. You’ll compile SME2-enabled pipelines with KleidiCV and KleidiAI, then run background blur, low-light enhancement, and temporal denoising on sample inputs. Finally, you’ll run the applications and benchmark binaries, and verify that expected output images are generated.

Frequently asked questions

AI-assisted

These FAQs were drafted with an approved AI-assisted workflow and reviewed by Arm contributors before publication. Human technical review remains part of the process so the final page reflects engineering rigor, accuracy, and Arm editorial standards.

Close
?
Where should I run the Docker build and which files does it use?
Run docker build from the repository root, specifying -f docker/Dockerfile with the build context set to docker/. The build uses the provided Dockerfile and scripts under the docker/ directory.
How do I fetch the large files after cloning the repository?
From inside the cloned repository, run git lfs install followed by git lfs pull. The commands download the required large assets referenced by Git LFS.
Where do I build the pipelines and where are the binaries placed?
Start a shell in the Docker container and build the pipelines there. You can find the compiled executables in the project’s bin/ directory.
What result should I expect after running the background blur pipeline?
The pipeline reads the specified input image and writes a transformed image to the output path you provide, for example test_output_cinematic_mode.png. Verify that the output file is created without errors.
How do I use the benchmark mode and what indicates it worked?
Run the provided benchmark executables in bin/: cinematic_mode_benchmark, low_light_image_enhancement_benchmark, and neural_denoiser_temporal_benchmark_4K. The executables run the core loop multiple times. A successful run completes without errors and allows you to observe behavior on your system.
Next