Demo

System Setting

After booting up, StarFive recommends to turn off CPU auto frequency regulation and run the following command under root:

$ su -
$ echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor

Enable HailoRT Monitor (optional)

Run the following command to enable Hailo monitor, which can monitor the utilization rate of NPU devices, models used, and FPS output in real-time:
```
$ hailortcli monitor
```
Figure 1. Enable Hailo Monitor
Open another shell (this shell must be the one running the demo program), and specify the environment variables:
```
$ export HAILO_MONITOR=1
```
Figure 2. Specify environment variables
When executing the demonstration case, HailoRT detects that the HAILO.MONITOR variable of this shell process is 1, and will output real-time monitoring data of the NPU module:
Figure 3. Monitor Data

Tappas Demo

The following demo cases can be directly used under Tappas.

Note: The demo under Tappas uses GStreamer videosink with xvimagesink and ximagesink, while VisionFive 2's Debian uses Wayland protocol, these demos need to modify their startup scripts by changing the ximagesink in video_stink_ element to waylandink:

Otherwise, the following error will occur during runtime:

Instance_segmentation:

$ cd /home/user/tappas/apps/h8/gstreamer/vf2/instance_segmentation/
$ ./instance_segmentation.sh -i /dev/video4

Note: The command /dev/video4 specifies the used USB camera.

Detection:

$ cd /home/user/tappas/apps/h8/gstreamer/vf2/instance_segmentation/
$ ./instance_segmentation.sh -i /dev/ video4-- video4

Note: Specify the use of YOLOv5 model through -network.

Cascading_networks:
```
$ cd /home/user/tappas/apps/h8/gstreamer/vf2/cascading_networks/
```
- object_detection_and_pose_estimation:
```
$ ./object_detection_and_pose_estimation.sh -i /dev/video4
```
  Note: When the following error occurs, check if the TAPPAS-WORKSPACE environment variable is missing.
  Figure 6. Error
  
  If this error occurs, you can set the environment variable: export TAPPAS-WORKSPACE=path_to-tappas/tappas/, where path_to-tappas is the path where the Tappas source code directory is located.

NpuDetectorLib Demo

Run the following command to download and compile NpuDetectorLib:

$ tar -xvf NpuDetectorLib.tar
$ cd NpuDetectorLib
$ cmake -H. -Bbuild -DSHOW_LABEL=ON -DBUILD_TESTER=ON
$ cmake --build build

Download the required resources:

$ wget -O models/yolov8s_nms.hef https://hailo-model-zoo.s3.eu-west-2.amazonaws.com/ModelZoo/Compiled/v2.13.0/hailo8l/yolov8s.hef

Run the following command to demo target recognition:

$ ./build/tests/TestExecutable -i /dev/video4 -m models/yolov8s_nms.json -a yolov8_nms