Workflow for Deep Learning Code Generation with MATLAB Coder

With MATLAB^® Coder™, you can generate code for prediction from a pretrained convolutional neural network (CNN), targeting an embedded platform that uses an Intel^® processor or an ARM^® processor. The generated code calls the Intel MKL-DNN or ARM Compute Library to apply high performance.

You can also use MATLAB Coder to generate generic C or C++ code for deep learning networks. Such C or C++ code does not depend on any third-party libraries.

1. Get a trained network by using Deep Learning Toolbox™. Construct and train the network or use a pretrained network. For more information, see:

   • Deep Learning in MATLAB (Deep Learning Toolbox).

   • Pretrained Deep Neural Networks (Deep Learning Toolbox).

   The network must be supported for code generation. See Networks and Layers Supported for Code Generation.

2. Load a network object from the trained network.

   See Load Pretrained Networks for Code Generation.

3. Generate C++ code for the trained network by using codegen or the MATLAB Coder app. See:

   • Code Generation for Deep Learning Networks with MKL-DNN

   • Code Generation for Deep Learning Networks with ARM Compute Library

   • Generate Generic C/C++ Code for Deep Learning Networks

Related Topics

• Deep Learning in MATLAB (Deep Learning Toolbox)
• Learn About Convolutional Neural Networks (Deep Learning Toolbox)
• Prerequisites for Deep Learning with MATLAB Coder
• Code Generation for Deep Learning Networks with MKL-DNN
• Deep Learning Code Generation on Intel Targets for Different Batch Sizes
• Code Generation for Deep Learning Networks with ARM Compute Library
• Code Generation for Deep Learning on ARM Targets
• Deep Learning Prediction with ARM Compute Using codegen
• Generate Generic C/C++ Code for Deep Learning Networks
• Deep Learning with GPU Coder (GPU Coder)