BNN with custom neural network Ultra96v2

Support
1

Hello everyone,

I am a beginner in ML on FPGA. I rebuilt bnn project and opened it. I used Vivado 2018.2.

I saw BlackBoxJam IP core. Can I modify this IP block if I want to use I new neural network?

Is it a black box for me? How I can modify it (from HLS or from Vivado)?

Can anyone explain to me what the steps are to get started?

2

You would be best checking on the BNN repo first: GitHub - Xilinx/BNN-PYNQ: Quantized Neural Networks (QNNs) on PYNQ
and corresponding docs for FINN project.
FINN | finn

Cathal

3

I suppose you followed these steps GitHub - Xilinx/BNN-PYNQ: Quantized Neural Networks (QNNs) on PYNQ

The BNN builds on top of GitHub - Xilinx/finn-hlslib: Vitis HLS library for FINN, and the function that includes the different layers are here BNN-PYNQ/bnn/src/network at master · Xilinx/BNN-PYNQ · GitHub.

As Cathal mention, you can find more documentation on BNN and FINN repos.