A selection of notebook examples are shown below that are included in the PYNQ image. The notebooks contain live code, and generated output from the code can be saved in the notebook. Notebooks can be viewed as webpages, or opened on a Pynq enabled board where the code cells in a notebook can be executed.
FPGA-based neural network inference for DAC 2018 contest
Accelerated OpenCV image filtering library.
FPGA-based neural network inference project
1st place in the DAC 2018 design contest for neural network object detection
Control of robotic car from PYNQ
PYNQ LED cube
Fudan University, Xilinx China
Controlling an LED cube from PYNQ
Hardware accelerated videoprocessing
Hardware accelerated compression
Quantised neural network
NTNU, University Sydney, Xilinx labs
Binarised neural network
Xilinx ISM, Trenz electronics
Industrial motor control
Overlay with network analysis capability
Quantized LSTM on PYNQ
Video filters with PR
Video filtering with partial reconfiguration
PYNQ computer vision
Build a vision processing pipeline from xfOpenCV
GZip on PYNQ
GZip compression with DEFLATE-compatible data, and fixed Huffman coding
Apache Spark on PYNQ
FIR filter example
Example of integrating a FIR filter
CNN on PYNQ
Imperial College London
HDMI Video processing
Ruhr University Bochum
Soft GPU on PYNQ
Extended Kalman filter
PYNQ community projects
A selection of projects from the PYNQ community is shown below. Note that some examples are built on different versions of the PYNQ image.
PYNQ has been widely used for machine learning research and prototyping.
FINN, an experimental framework from Xilinx Research Labs to explore deep neural network inference on FPGAs. It specifically targets quantized neural networks, with emphasis on generating dataflow-style architectures customized for each network.
FINN makes extensive use of PYNQ as a prototyping platform.