PROJECT NAME:

PYNQ Classification - Python on Zynq FPGA for Convolutional Neural Networks (Alpha Release)

BRIEF DESCRIPTION:

This repository presents a fast prototyping framework, which is an Open Source framework designed to enable fast deployment of embedded Convolutional Neural Network (CNN) applications on PYNQ platforms.

INSTRUCTIONS TO BUILD AND TEST THE PROJECT:

Repository Organisation

The project demo accepts pre-trained CNN models in either Caffe or Theano syntax, hence the step 1 and 2 introduces how to install Caffe and Theano (with Lasagne) on PYNQ. Step 3 explains how to download and run the demos for LeNet and CIFAR-10 (Caffe "quick" version) models.

For a quick overview on the project please watch my video tutorial.

1. PYNQ SD Card Image

We have prepared a SD card image with pre-installed Caffe and Theano dependencies. A SD card with at least 16GB is needed. The static IP for the PYNQ Jupyter Notebook is 192.168.2.99

Download Link (on Baidu Drive)

If you wish to setup Caffe and Theano dependencies on your own, please see MANUAL_INSTAL.md for instructions. (NOT RECOMMENDED since multiple issues have been reported)

2. Vivado Project Setup - How to implement more CNN models?

The VIVADO_SIDE.7z package contains the Vivado project needed to generate bitstreams for customised CNN models. It contains the "layer IP library", "base project" and "block design project" as mentioned in the tutorial video.

References

1. BNN-PYNQ

If you find BNN-PYNQ useful, please cite the FINN paper:

@inproceedings{finn,
author = {Umuroglu, Yaman and Fraser, Nicholas J. and Gambardella, Giulio and Blott, Michaela and Leong, Philip and Jahre, Magnus and Vissers, Kees},
title = {FINN: A Framework for Fast, Scalable Binarized Neural Network Inference},
booktitle = {Proceedings of the 2017 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays},
series = {FPGA '17},
year = {2017},
pages = {65--74},
publisher = {ACM}
}

2. Caffe

Caffe is released under the BSD 2-Clause license. The BAIR/BVLC reference models are released for unrestricted use.

Please cite Caffe in your publications if it helps your research:

@article{jia2014caffe,
  Author = {Jia, Yangqing and Shelhamer, Evan and Donahue, Jeff and Karayev, Sergey and Long, Jonathan and Girshick, Ross and Guadarrama, Sergio and Darrell, Trevor},
  Journal = {arXiv preprint arXiv:1408.5093},
  Title = {Caffe: Convolutional Architecture for Fast Feature Embedding},
  Year = {2014}
}

3. Theano

@ARTICLE{2016arXiv160502688short,
   author = {{Theano Development Team}},
    title = "{Theano: A {Python} framework for fast computation of mathematical expressions}",
  journal = {arXiv e-prints},
   volume = {abs/1605.02688},
 primaryClass = "cs.SC",
 keywords = {Computer Science - Symbolic Computation, Computer Science - Learning, Computer Science - Mathematical Software},
     year = 2016,
    month = may,
      url = {http://arxiv.org/abs/1605.02688},
}

4. Lasagne

@misc{lasagne,
  author       = {Sander Dieleman and
                  Jan Schlüter and
                  Colin Raffel and
                  Eben Olson and
                  Søren Kaae Sønderby and
                  Daniel Nouri and
                  others},
  title        = {Lasagne: First release.},
  month        = aug,
  year         = 2015,
  doi          = {10.5281/zenodo.27878},
  url          = {http://dx.doi.org/10.5281/zenodo.27878}
}