# PYNQ + Vitis AI Library

**URL:** <https://discuss.pynq.io/t/pynq-vitis-ai-library/1178>\
**Category:** Support\
**Created:** [April 25, 2020, 3:26pm UTC](https://discuss.pynq.io/t/pynq-vitis-ai-library/1178 "2020-04-25T15:26:14Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![John](https://avatars.discourse-cdn.com/v4/letter/j/a3d4f5/32.png) [@John](https://discuss.pynq.io/u/John)\
**Post date:** [April 25, 2020, 3:26pm UTC](https://discuss.pynq.io/t/pynq-vitis-ai-library/1178/1 "2020-04-25T15:26:14Z")

</div>

Hello everyone,

I was wondering if it is possible to use the Vitis AI library located [here](https://github.com/Xilinx/Vitis-AI/tree/master/Vitis-AI-Library) togheter with PYNQ Z1 or Z2 board? In particular I was focusing on the [face detect](https://github.com/Xilinx/Vitis-AI/tree/master/Vitis-AI-Library/facedetect) one.

Thank you very much!

---

<div class="post-metadata">

**Author:** ![rock](https://sea1.discourse-cdn.com/flex019/user_avatar/discuss.pynq.io/rock/32/32_2.png) [@rock](https://discuss.pynq.io/u/rock)\
**Post date:** [April 27, 2020, 3:57pm UTC](https://discuss.pynq.io/t/pynq-vitis-ai-library/1178/2 "2020-04-27T15:57:52Z")

</div>

Not really at this moment. The vivado flow works with 2019.1 tools, but for vitis AI, the main issue is the DPU released is not supporting zynq7000 for now.

---

<div class="post-metadata">

**Author:** ![John](https://avatars.discourse-cdn.com/v4/letter/j/a3d4f5/32.png) [@John](https://discuss.pynq.io/u/John)\
**Post date:** [April 27, 2020, 4:51pm UTC](https://discuss.pynq.io/t/pynq-vitis-ai-library/1178/3 "2020-04-27T16:51:46Z")

</div>

Thank you for clarifying,

is there any way to quantize neural networks with vivado 2019.1 in order to create a pynq overlay?

---

<div class="post-metadata">

**Author:** ![rock](https://sea1.discourse-cdn.com/flex019/user_avatar/discuss.pynq.io/rock/32/32_2.png) [@rock](https://discuss.pynq.io/u/rock)\
**Post date:** [April 27, 2020, 7:47pm UTC](https://discuss.pynq.io/t/pynq-vitis-ai-library/1178/4 "2020-04-27T19:47:02Z")

</div>

I think there are some Xilinx documents that you can find online. It requires you to build a dpu.ko kernel module and put that module on the board. For quantization please refer to the DNNDK user guide.

---

<div class="post-metadata">

**Author:** ![John](https://avatars.discourse-cdn.com/v4/letter/j/a3d4f5/32.png) [@John](https://discuss.pynq.io/u/John)\
**Post date:** [April 29, 2020, 9:34am UTC](https://discuss.pynq.io/t/pynq-vitis-ai-library/1178/5 "2020-04-29T09:34:51Z")

</div>

Thanks for your answer! I tried to find some Xilinx documents online related to the dpu.ko module and how it could work with the PYNQ image but I didn’t find anything but [GitHub - louisliuwei/pynq-dpu: Migrate Xilinx edge AI solution to PYNQ](https://github.com/louisliuwei/pynq-dpu) or [dnndk-pynqz2/build-pynqz2-system.md at master · wutianze/dnndk-pynqz2 · GitHub](https://github.com/wutianze/dnndk-pynqz2/blob/master/build-pynqz2-system.md).  
Is this approach compatible with PYNQ 2.5 ? Is there any other Xilinx documents that I’m missing? Thank you very much in advance