search

Xilinx / Vitis-AI-Tutorials

MIT License
380 stars 145 forks source link

How to run the compiled files into target board; tar, xmodel #115

Open ElonGUI opened 9 months ago

ElonGUI commented 9 months ago

Keyword：Compiled, tar, Optimizer, MobileNetV2 , VA3.5 I want run the example of MobileNetV2 in ZCU104. I want to tar the compiled file(.xmodel) and transfer it into my board(zcu104). But I can't find zcu104_build_* when I run the tar command in the tutorial, as shown below.

tmp

Later in 5.5.2, I find that <targetboard>_build_*.tar is included in dogs-vs-cats_mobilenetv2.tar， but I don't know where <targetboard>_build_*.tar or zcu104_build_*.tar is generated from. I looked at the run_all_mobilenetv2_target.sh and found that I didn't have build_get_dpu_fps.sh either. I tried to run through the instructions in the tutorial from the beginning, but with no results. How should this situation be resolved? Thank you very much~ （By the way, the author, Daniele Bagni, is very carefully to describe in this chapter, avoiding giving beginners lots of problems, Thank you very much. But would someone like to add how to burn the generated model into the target development board?）

ElonGUI commented 9 months ago

I'm here to answer my own question. After completing Step 7 of MobileNetV2 and Cifar, I learned that I needed to run script run_pruning.sh to produce the tar file. Then cope the tar into target board. I successfully ran the script on my ZCU104，and I‘m trying to analyze the log of this result. Thanks for the author, Daniele Bagni, again. It is verig detailed tutorial.