Hello,
Im using AI-G and test ok with Pi camera v1. then i try RTPM to display to my laptop.
But i cannot run main.py in RTPM. does it can run in ubuntu?
Hello, this is the TOPST manager.
RTPM can run on Ubuntu.
The current error occurs because the library does not exist, so you need to link libvisionprotocol.so under the /usr/lib directory.
It can be executed with the following command:
$ sudo ln -s /home/ead/project/rtpm/model/visionprotocol/cython_visionprotocol/libvisionprotocol.so.2.0.0 /usr/lib/libvisionprotocol.so.2
Thank you.
Hello,
Thanks for reply to my question. But I get a new error both on window10 and ubuntu.
On Window, I stuck when start rtpm. it “waiting for client connection…”. then I try to nc 192.168.1.202 to connect AI-G with pc and stuck at Authentication again.
in 2.4.2 Open RTPM Application. Step 1: Copy the rtpm folder (\rtpm) from the SDK directory to the Host PC (Windows PC). but I don’t find rtpm folder
On Ubuntu, I pass last error and get another one. It seems like Qt platform doesn’t install even it was found. I try install all library it needed but still error.
Thank you.
Hello.
Rtpm is basically designed to communicate through the IP 192.168.0.8.
Please refer to section 4.2.2 on the page TOPST and set the Ethernet IP to 192.168.0.8.
The ai-g board is by default set to the IP 192.168.0.100.
If you follow the settings on that page, you will be able to use rtpm immediately.
For errors occurring on Ubuntu, please install llvm and set the QT platform to eglfs to run it successfully.
$ sudo apt install llvm
$ export QT_QPA_PLATFORM=wayland
$ python3 main.py
Thank you.
Hello,
Thanks for reply.
I set Ethernet IP 192.168.0.8. but the ai-g board is set IP 192.168.0.95 (command “ip a” in board). I config IP in section 5.1.2 on the page TOPST.
Can I change IP communication of RTPM and ai-g
On ubuntu I follow your instructions, rtpm ‘s window is opened but I cannot do anything. It looks like plugin input is missing or freeze too. I try some input, but nothing happens.
export QT_QPA_EGLFS_INPUT=evdev
sudo usermod -a -G input cafnar
Thank you.
Hello.
On window, I try to config but it still not communication.
On ubuntu, I got this error below after try export QT_QPA_PLATFORM=xcb.
(rtpm) cafnar@cafnar-Legion-Y7000P-IRH8:~/Downloads/rtpm$ QT_QPA_PLATFORM=xcb cd /home/cafnar/Downloads/rtpm ; /usr/bin/env /home/cafnar/miniconda3/envs/rtpm/bin/python /home/cafnar/.vscode/extensions/ms-python.debugpy-2025.14.1-linux-x64/bundled/libs/debugpy/adapter/../../debugpy/launcher 41919 – /home/cafnar/Downloads/rtpm/main.py
Traceback (most recent call last):
File “/home/cafnar/miniconda3/envs/rtpm/lib/python3.10/runpy.py”, line 196, in _run_module_as_main
return \_run_code(code, main_globals, None,
File “/home/cafnar/miniconda3/envs/rtpm/lib/python3.10/runpy.py”, line 86, in _run_code
exec(code, run_globals)
File “/home/cafnar/.vscode/extensions/ms-python.debugpy-2025.14.1-linux-x64/bundled/libs/debugpy/adapter/../../debugpy/launcher/_main_.py”, line 92, in
main()
File “/home/cafnar/.vscode/extensions/ms-python.debugpy-2025.14.1-linux-x64/bundled/libs/debugpy/adapter/../../debugpy/launcher/_main_.py”, line 48, in main
launcher.connect(host, port)
File “/home/cafnar/.vscode/extensions/ms-python.debugpy-2025.14.1-linux-x64/bundled/libs/debugpy/adapter/../../debugpy/launcher/../../debugpy/launcher/_init_.py”, line 28, in connect
sock.connect((host, port))
ConnectionRefusedError: [Errno 111] Connection refused
I give up on RTPM and buy a new 5 inch DSI. But i got the new error. the display just blink 1 time when I power up AI-G. I run my model and the display still off.
root@ai-g-topst:~# dmesg | grep -i display
[ 0.033943] [INF][VIOC_CONFIG] disabled [this is mandatory for vin display]
[ 1.360293] [INF][OVERLAY] display_num = 0
[ 1.360603] [INF][OVERLAY] display_num = 0
[ 1.360831] [INF][OVERLAY] display_num = 0
$ tcnnapp -p /dev/video0 -W 800 -H 480 -w 1920 -h 1080 -n mb2-ssd-lite_quantized/
Hello.
It appears that the RTPM errors currently occurring on Windows and Ubuntu are due to IP configuration issues.
First, please check the communication status on the board using the ping 192.168.0.8 command.
For the display, could you please confirm that the TOPST AI logo appears on the display when the board is powered on?
Also, please let us know which display device you are using!
Thank you.
Hello.
I use 5 inch DSI LCD waveshare and No TOPST AI logo appears. it just blink in 1 frame.
When i check $ dmesg | grep mipi. It just show csi and no dsi. Does it need driver? if yes, how can i do?
Thank you.
Hello.
First, please check the display connection cable.
If the same problem occurs afterward, it seems that the driver has not been ported.I am sharing the information of the display that was tested.
https://a.aliexpress.com/_c3hEegrN
Thank you.
Hello.
I connect display with pi 3 model b and it works. but when connect to AI-G. it doesn’t work.
I swap 2 cable of camera with display and run the same command $ tcnnapp -n mb2-ssd-lite_quantized/.
I find out that white cable connect to camera, camera work (terminal show 1 object is detected
x1 y1 x2 y2 = 1.8407 1.8407 298.1592 298.1592 class: 12 score: 55.1757) . but brown cable connect to camera, it doesn’t work (it stucks Demo : tc-nn-app Demo) . however. display doesn’t work no matter what cable it connect.
Thank you.
Hello.
Can you share me the driver or guide for display?
Thank you.
First, we only conducted tests using the DSI display I sent you above.
Could you possibly send me the link to the product you used?
Thank you.
Hello.
I send you link information of my display.
I have the wrong display but I think elecrow and waveshare is kinda similar.
Thank you.
Hello.
At my lab, IP is 192.168.1. *and I cannot change. My question above mean Can i modify IP of board and host PC to 192.168.1.*100 and 192.168.1.8? if yes, does it work with rtpm?
Thank you.
Hello
First of all, instead of modifying the IP of the Ethernet port of the lab environment itself, you are modifying the IP of the Ethernet port that connects the lab desktop and AI-G, so it would be good to double-check this.
If you want to use RTPM with 192.168.1.XX, you need to modify the tc-nn-app source.Inside the Yocto SDK, under the directory “topst-sdk/build/ai-g-topst/tmp/work/cortexa53-telechips-linux/tc-nn-app/1.0.0-r0/git/src”
you can modify DEFAULT_TARGET_IP_ADDRESS in the NnAppMain.c source according to the current desktop Ethernet.
Thank you.
Hello.
Firstly, do I need re-image firmware after modify the tc-nn-app source ?
Secondly, it seem like NnAppMain.c in your image is on AI-G but I cannot find it
and in the path on desktop I must git clone GitHub - topst-development/tc-nn-app: tc-nn-app for TOPST SBC .
Finally, in session 2.4.2 Open RTPM Application, step 1. Copy the rtpm folder (\rtpm) from the SDK directory to the Host PC (Windows PC). however, I don’t see any rtpm in folder topst-SDK.
Thank you.
Hello.
When downloading the AI-G firmware, you need to modify the source in the Yocto SDK environment you built.
After completing the process up to 3.7 on the relevant page, instead of running the command
bitbake telechips-topst-ai-image,
you only need to build bitbake -c compile -f tc-nn-app to generate the tcnnapp source.
Then, open NnAppMain.c, modify the source, rebuild tcnnapp, and the generated tcnnapp under topst-sdk/build/ai-g-topst/tmp/work/cortexa53-telechips-linux/tc-nn-app/1.0.0-r0/build can be transferred to the AI-G board for use.
Thank you.
Hello.
When I run your command. i got this error bitbake -c compile -f tc-nn-app.
I check nano /home/dmin/topst-sdk/poky/meta-topst-bsp/recipes-kernel/linux/linux-topst_5.10.bb and find out it have 2 version of linux 5.10.205 and 5.10.223.
Does it still work?
Thank you.
Hello,
If you run the command source poke/meta-topst/build-topst.sh and select a machine, it is set to automatically build kernel 5.10.223. Kernel 5.10.205 is for D3-G.
In my environment, when I followed the procedure as instructed in the Docs, it built without any issues.
Could you tell me about the environment in which you built it?
I recommend checking your environment settings and related package installations once more and then rebuilding!
Hello.
I build on both ubuntu 22.04 and window 10 pro. In run command source poke/meta-topst/build-topst.sh and select 1 (AI-G) successful. but when run command bitbake telechips-topst-ai-image or bitbake -c compile -f tc-nn-ap, the error above happen.
Thanks.