This tutorial tells to run the innodisk VVAS demo.
For runnning the innodisk VVAS demo, you will need three parts:
- VVAS
Vitis Video Analytics SDK. - Xilinx smartcam
We will use the library of Xilinx smartcam to draw the result of AI inference. - Innodisk demo
Contains example files for innodisk VVAS demo.
- VVAS 2.0 is default built-in with the system, check out this page if user needs to build VVAS manually.
- This demo will use the library of Xilinx smartcam and aibox-reid to draw the result of AI inference and using the preprocessing IP.
- Following process is using smartcam as example, remember to install aibox-reid as well.
- Preparation Petalinux sdk including Vitis ai, opencv (over 4.4), jansson.
-
Download source code.
git clone https://github.com/Xilinx/smartcam.git cd smartcam
-
Source petalinux sdk.
unset LD_LIBRARY_PATH source <path-to-sdk>/environment-setup-aarch64-xilinx-linux
-
Build the project.
chmod 755 ./build.sh ./build.sh <path-to-sdk>
If shows error message as below:
ERROR: missing libcorrespondence.a
Follow this post of xilinx forum will fix the issue.
-
Copy the .rpm file from x86 host to the rootfs of k26.
-
Install the .rpm file on k26.
rpm -ivh --force smartcam-1.0.1-1.aarch64.rpm
- Preparation Install the VVAS and xilinx smartcam by following previous section.
rpm -ivh --force vvas_demo-0.1-1.aarch64.rpm
- Using the firmware
aibox-reid
which from kria-apps-firmware.
- Plug the HDMI port of carrier board to a screen.
- Disable the display in device-tree for video mixer on PL, avoiding video mixer initial fail that makes user still using DP on PS.
display@fd4a0000 { status = "disabled"; }; &zynqmp_dp_snd_codec0 { status = "disabled"; }; &zynqmp_dp_snd_pcm0 { status = "disabled"; }; &zynqmp_dp_snd_pcm1 { status = "disabled"; }; &zynqmp_dp_snd_card { status = "disabled"; };
-
Update FPGA application.
xmutil unloadapp xmutil loadapp kv260-aibox-reid
-
Excute the script from Innodisk demo.
cd ~/demo # basic VVAS demo sudo ./run_demo_vvas.sh # VVAS demo including uvc camera with face detection sudo ./run_demo_vvas_cam.sh