cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Observer
Observer
137 Views
Registered: ‎06-05-2019

How to make DPU application for AI-Model-Zoo

Dear Xilinx,

I want to execute object detection (such as SSD or Yolo) AI application on my custom FPGA.
I found some object detection sample models in Xilinx/AI-Model-Zoo <https://github.com/Xilinx/AI-Model-Zoo>, and fortunately that include quantized models (deploy.caffemodel or deploy_model.pb).
So I compiled the model with DNNC and suceeded to generate .elf file in dnnc_output folder.

Then my question is how can I make DPU executable application for my FPGA?
My understanding is we need an application source (such as main.cc) and some include files besides .elf file to compile the application, but is it correct?
If I'm correct, is there any available sample source files to compile the appilcation for AI-Model-Zoo?

Now I'm thinking to use "tf_ssdmobilenetv2_coco_300_300" and "tf_yolov3_voc_416_416" for my test, so I would appreciate if you could provide the samples for these models.
It's just for my internal test to evaluate DPU performance so no need for validations or warranties of sample files.


FYI my FPGA design details are as follows.

Board : ZCU104
DPU arch : B1024 x1, B2304 x1, etc
DPU TRD ver : dpu-trd-2019-1
Vivado ver : 2019.1
Petalinux ver : 2019.1

Looking forward to hearing from you and thanks for your support.