MATLAB Coder Interface for Deep Learning (original) (raw)

Use MATLAB Coder to generate C and C++ code for deep learning networks

4.4K Downloads

Updated11 Dec 2024

• Share

• Overview

• Reviews (8)

• Discussions (7)

MATLAB Coder generates C and C++ code from MATLAB code for a variety of hardware platforms, from desktop systems to embedded hardware. It supports most of the MATLAB language and a wide range of toolboxes, and you can deploy a variety of pretrained deep learning networks such as YOLOv2, ResNet-50, SqueezeNet, and MobileNet from Deep Learning Toolbox. You can generate optimized code for pre-processing and post-processing along with your trained deep learning networks to deploy complete applications.

• Supported deep learning networks

With MATLAB Coder or Simulink Coder, MATLAB Coder Interface for Deep Learning provides the ability to generate plain (library-independent) C/C++ code for deep learning networks. Code replacement libraries can be used to incorporate processor-specific intrinsics for the target hardware (e.g. ARM Cortex-A/M processors). Additionally, it provides the option to generate code that calls into the following target-specific, optimized libraries:

• Intel oneAPI Deep Neural Network Library (oneDNN, formerly MKL-DNN): For Intel CPUs that support AVX2
• ARM Compute Library: For ARM Cortex-A processors that support NEON instructions

When used in Simulink with Deep Learning Toolbox and without MATLAB Coder or Simulink Coder, you can accelerate simulations of Simulink models that include deep learning blocks using the Intel oneDNN optimization library.

For more information on building supported optimization libraries, please see these links:

• MATLAB Coder: How do I build the Intel MKL-DNN library for Deep Learning C++ code generation and deployment?
• MATLAB Coder: How do I build the ARM Compute Library for Deep Learning C++ code generation and deployment?

To learn more about the recommended settings for optimizing the inference perfomance of plain, library-independent C/C++ code generated from deep learning networks, please see the below link:

• How can I optimize the performance of library-independent C/C++ code generated from deep learning networks?

This support package is functional for R2018b and beyond.