Prototype Deep Learning Networks on FPGA and SoCs Workflow

To prototype and deploy your custom series deep learning network, create an object of class dlhdl.Workflow. Use this object to accomplish tasks such as:

• Compile and deploy the deep learning network on specified target FPGA or SoC board by using the deploy function.

• Estimate the speed of the deep learning network in terms of number of cycles by using the estimate function.

• Execute the deployed deep learning network and predict the classification of input images by using the predict function.

• Calculate the speed and profile of the deployed deep learning network by using the predict function. Set the Profile parameter to on.

This figure illustrates the workflow to deploy your deep learning network to the FPGA boards.