Skip to content

ZFTurbo/Verilog-Generator-of-Neural-Net-Digit-Detector-for-FPGA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

57 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Verilog Generator of Neural Net Digit Detector for FPGA

It's the project which train neural net to detect dark digits on light background. Then neural net converted to verilog HDL representation using several techniques to reduce needed resources on FPGA and increase speed of processing. Code is production ready to use in real device. It can be easily extended to be used with detection of other objects with different neural net structure.

Requirements

Python 3.5, Tensorflow 1.4.0, Keras 2.1.3

How to run:

  • python r01_train_neural_net_and_prepare_initial_weights.py
  • python r02_rescale_weights_to_use_fixed_point_representation.py
  • python r03_find_optimal_bit_for_weights.py
  • python r04_verilog_generator_grayscale_file.py
  • python r05_verilog_generator_neural_net_structure.py

Verilog already added in repository in ''verilog'' folder. It has everything you need including all code to interact with camera or screen. Neural net verilog description is located in ''verliog/code/neuroset'' folder.

Neural net structure

Neural Net Structure

Device

To recreate the device you need 3 components:

Connection of components

Connection scheme

De0-Nano connection

Connection photo

Demo video with detection

Convolutional Neural Net implementation in FPGA (Demo)

Notes

  • You can change constant num_conv = 2 in r05_verilog_generator_neural_net_structure.py to 1, 2 or 4 convolutional blocks which will work in parallel. More blocks will require more LE in FPGA, but increase the overall speed.

  • Comparison table for different bit weights and number of convolution blocks below (red rows: unable to synthesize, due to Cyclone IV limitations).

Used FPGA resources

Related project

The similar project but with more complicated and widely used neural net: MobileNet (v1). It uses some other set of devices. It has similar code structure. It has fast speed (>40 FPS) and much better accuracy comparing to this project. It suitable for most image classification tasks in real time.

Citation

You can find detailed description of the method in our paper (or unpaywalled preprint). If you find this work useful, please consider citing:

  @inproceedings{solovyev2019fixed,
    title={Fixed-point convolutional neural network for real-time video processing in FPGA},
    author={Solovyev, Roman and Kustov, Alexander and Telpukhov, Dmitry and Rukhlov, Vladimir and Kalinin, Alexandr},
    booktitle={2019 IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering (EIConRus)},
    pages={1605--1611},
    year={2019},
    organization={IEEE}
  }