Skip to content
/ spooNN Public
forked from fpgasystems/spooNN

FPGA-based neural network inference project with an end-to-end approach (from training to implementation to deployment)

License

Notifications You must be signed in to change notification settings

vxgu86/spooNN

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

spooNN

picture

This is a repository for FPGA-based neural network inference, that delivered the highest FPS in the international contest for object detection as part of Design Automation Conference 2018 (https://www.dac.com/content/2018-system-design-contest). The contents of spooNN enable an end-to-end capability to perform inference on FPGAs; starting from training scripts using Tensorflow to deployment on hardware (PYNQ http://www.pynq.io/).

picture The final rankings are published at http://www.cse.cuhk.edu.hk/~byu/2018-DAC-HDC/ranking.html

Repo organization

  • hls-nn-lib: A neural network inference library implemented in C for Vivado High Level Synthesis (HLS).
  • mnist-cnn: helloworld project, showing an end-to-end flow (training, implementation, FPGA deployment) for MNIST handwritted digit classification with a convolutional neural network.
  • halfsqueezenet: The object detection network, that ranked second in DAC 2018 contest, delivering the highest FPS at lowest power consumption for object detection.

About

FPGA-based neural network inference project with an end-to-end approach (from training to implementation to deployment)

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 63.4%
  • Python 19.8%
  • Tcl 6.8%
  • Objective-C 5.4%
  • C++ 3.6%
  • C 0.7%
  • Other 0.3%