├── jupyter // Pynq application
│ └── filter2d_demo.ipynb
├── overlay // Bitstream and hardware handoff
│ ├── filter2d.bit
│ └── filter2d.hwh
└── src // HLS design and testbench
├── dst.txt
├── filter2d.cpp
├── filter2d.h
├── filter2d_test.cpp
├── src.txt
└── tcl_script.tcl
A HLS implementation of 2D convolution algorithm for custom 3*3 kernel and images smaller than 128*128, no padding.
This project was presented at Xilinx Summer Camp 2022.
-
Platform: PYNQ-Z2, Filter2D IP running in 150MHz.
-
Python requirements: opencv-python, matplotlib, plotly, pynq
Demo(硬件自定义二维卷积IP核实现图片浮雕效果): Here
Some results:
