KerasPreprocessings

KerasPreprocessings is a part of the DeepHLS toolchain.

Citing us

If you find this work helpful for your research, please consider citing our paper published in The 27th IEEE International Conference on Electronics, Circuits and Systems (ICECS), 2020.

Riazati, Mohammad, Masoud Daneshtalab, Mikael Sjödin, and Björn Lisper. "DeepHLS: A complete toolchain for automatic synthesis of deep neural networks to FPGA." In The 27th IEEE International Conference on Electronics, Circuits and Systems (ICECS), pp. 1-4. IEEE, 2020.

Description

For more information about the toolchain and how to use KerasPreprocessings, please refer to DeepHLS tool opensourced under the same github user page.

Required packages

• numpy
• tensorflow
• keras
• tensorflow_datasets
• pydot

How to use

• Clone the repository.
• Create the "outputs" directory just beside KerasPreprocessings.py
• Determine the network and the dataset in the beginnig of the code
• Run
• Use output_arch.py as the input for DeepHLS to create the synthesizable C implementation
• Copy data.h and param.h to DeepSimulator folder as the input dataset and parameters.
• (Copy the file created by DeepHLS to DeepSimulator folder as the source C code for simulation)