8000 Pre-Post Processing · Xilinx finn · Discussion #1303 · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

Pre-Post Processing #1303

Closed Answered by auphelia
m7md5303 asked this question in Q&A
Pre-Post Processing #1303
@m7md5303 m7md5303
Mar 22, 2025 · 1 comments · 1 reply
Answered by auphelia Return to top
Discussion options

m7md5303
Mar 22, 2025

You must be logged in to vote
Answered by auphelia Apr 3, 2025

Hi @m7md5303 ,

Thanks for your questions!

Regarding question 1:
You can find more information on this in the tfc end2end example, here: https://github.com/Xilinx/finn/blob/main/notebooks/end2end_example/bnn-pynq/tfc_end2end_example.ipynb
In the section about Adding Pre- and Postprocessing:

Brevitas preprocesses BNN-PYNQ network inputs with torchvision.transforms.ToTensor() prior to training, which converts 8-bit RGB values into floats between 0 and 1 by dividing the input by 255. We can achieve the same effect in FINN by exporting a single-node ONNX graph for division by 255 (which already exists as finn.util.pytorch.ToTensor and merging this with our original model. Finally, we're going…

View full answer

Replies: 1 comment 1 reply

Comment options

auphelia
Apr 3, 2025
Maintainer

You must be logged in to vote
1 reply
@m7md5303
Comment options

m7md5303 Apr 3, 2025
Author

Answer selected by auphelia
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Category
Q&A
Labels
None yet
2 participants
@m7md5303 @auphelia
0