[Feature]: add board of CanMV-k230-LP4 3.0 support

[wilesun]

Feature description

CanMV-k230-LP4 3.0 AI board has 1G Memory and wifi is RTL8189FV,please add it to boards list

Want resolve what problem

CanMV-k230-LP4 3.0 AI board has 1G Memory and wifi is RTL8189FV,please add it to boards list

Anything else

No response

[wangjianxin-canaan]

The current repository will be supported in the next version. the https://github.com/kendryte/canmv_k230 repository is already supported,the configs file is k230_canmv_v3p0_defconfig ,but the repository is only rtsmart system,image can download from https://github.com/kendryte/canmv_k230/releases/download/v1.1/CanMV-K230-V3P0_micropython_v1.1-0-g5a6fc54_nncase_v2.9.0.img.gz .

[wilesun5060]

How long does it take to release this version?I bought a board(CanMV-k230-LP4 3.0 AI board) but never used it.

[DingliZhang]

@wilesun5060 New SDK of v1.8.0 has released. Have you try to build image for the board(CanMV-k230-LP4 3.0 AI board)? I have the same problem as you.
Thanks a lot.

[briliantnugraha]

hi @DingliZhang @wilesun5060 @wangjianxin-canaan @wilesun has anyone successfully use this v3.0 LP4?
I can't connect it through CanMV IDE (tested with Linux canmv ide), Linux + RT OS SDK, Debian, nor Ubuntu @@

I've been stuck for 1-2 weeks now to no avail, please help @@

NOTE: FYI, I burn the image with balena etcher (Linux) or Rufus (windows).
UPDATE: problem solved, need to connect both type Cs to pc/laptop.

---

Btw, has anyone been able to run the AI model? For example, face_detection.py? I encounter this error "can not found sensor on 2" when running it.
FYI, this is a default setting, I've fixed some trivial path issues here and there (like /sdcard/app/test/ into /sdcard/examples/)

[wilesun5060]

Thank you! I have tested it on CanMV-k230-LP4 3.0. Two cores can run. I have tried run some demos (anything involving network communication, such as rtsp client and server, still cannot be used). Here are some test feedback:

1. Ping test on the small core, ping the host through eth0, Ethernet packet loss, and stuck for a while
2. Communication between the big core and the little core, or I don’t know how to use it, but I feel strange, why not use the virtual network device in the big core to shield these problems.
3. The big core gives me the feeling that it is too troublesome to develop on it, so I hope that k230_linux_sdk can be completed quickly and support V3. It would be better if these two projects can be merged into firmware.

[briliantnugraha]

Upvote for this feature @wangjianxin-canaan , please help us stranded LP V3.0 usersto make this RISC-V device version works! thanks
NOTE: meanwhile, I've been watching this repo as a potential alternative to the k230 SDK. HoweverB, I am a little worried (again) that my LP V3.0 device is not supported @@

[DingliZhang]

hi @DingliZhang @wilesun5060 @wangjianxin-canaan @wilesun has anyone successfully use this v3.0 LP4? I can't connect it through CanMV IDE (tested with Linux canmv ide), Linux + RT OS SDK, Debian, nor Ubuntu @@

I've been stuck for 1-2 weeks now to no avail, please help @@

NOTE: FYI, I burn the image with balena etcher (Linux) or Rufus (windows). UPDATE: problem solved, need to connect both type Cs to pc/laptop.

Btw, has anyone been able to run the AI model? For example, face_detection.py? I encounter this error "can not found sensor on 2" when running it. FYI, this is a default setting, I've fixed some trivial path issues here and there (like /sdcard/app/test/ into /sdcard/examples/)

Sorry I didn't make any further attempts. I'm waiting for officials to provide pre-built binary images.

[biti-en]

Good news! @DingliZhang @wilesun5060 @wangjianxin-canaan @wilesun
in v1.2, it seems that the micropython could work with the AI demo examples inside. But be aware that the otg+power (NOTE: in board, they only write "power") needs to be connected to at least USB3.0. I tried with USB2.0 cable, it stucks (maybe low wattage).

So far, I've tested face detection, object detection, and hand keypoint detection; and they are working, yay~
For other demo examples, it seems working fine. Below are some SS of my tests.

vid_1.mp4

vid_2.mp4

My current problem is that I don't know how to turn on the attached display that comes with my canmv k230 @@.

[wangjianxin-canaan]

My current problem is that I don't know how to turn on the attached display that comes with my canmv k230 @@.

Please refer to https://developer.canaan-creative.com/k230_canmv/zh/main/zh/example/media/display.html 。
ai demo please refer to https://developer.canaan-creative.com/k230_canmv/zh/main/zh/example/ai/AI_Demo%E8%AF%B4%E6%98%8E%E6%96%87%E6%A1%A3.html
，change hdmi to lcd,

[biti-en]

wow that's stupid of me not checking that. In fact, I've just tried and it works!

However @wangjianxin-canaan , the image seems to be rotated 180', and I can't find "rot90" function inside to rotate it.
Or maybe, is there any function in the pipeline to rotate it by 180' for lcd @wangjianxin-canaan ?

Note: I find it weird that in the frame buffer, the image is in correct orientation, but not correct in the LCD.

test.mp4

[wangyan1998]

wow that's stupid of me not checking that. In fact, I've just tried and it works!

However @wangjianxin-canaan , the image seems to be rotated 180', and I can't find "rot90" function inside to rotate it. Or maybe, is there any function in the pipeline to rotate it by 180' for lcd @wangjianxin-canaan ?

Note: I find it weird that in the frame buffer, the image is in correct orientation, but not correct in the LCD.

test.mp4

For a common camera demo, please refer to Sections 2.4 and 2.5 in the link to set the hmirror and vflip of the image.The link is as follows: https://developer.canaan-creative.com/k230_canmv/en/main/api/mpp/K230_CanMV_Sensor_Module_API_Manual.html.
If you are using the scripts in 05-AI-Demo, please set hmirror and vflip in pl.create(), such as pl.create(hmirror=True, vflip=True). If the image is still incorrect, you can try different combinations of True and False until the correct image orientation appears.

[briliantnugraha]

I've confirmed that adding hmirror=True, vflip=True will correct the orientation in the LCD, although the texts are 180' rotated. Thanks for the help @wangjianxin-canaan @wangyan1998 :)

[briliantnugraha]

However, I am wondering, could I run the code without it connected to the CanMV IDE @wangjianxin-canaan ?

FYI, I see google overview that I need to flash micropython to the device (canmv k230 in my case), connect to the device and run the code directly. Does it work the same with the current version of canmv k230's micropython?
FYI, my micropython version is 1.2.1 (released a few days ago). And I also try to connect to serial network (COM6 or COM7), and it is still unusable IMHO, but cmiiw

For other alternative, like Linux + RTOS is not yet available for V3.0 LP4 model @@

[wangjianxin-canaan]

https://developer.canaan-creative.com/answer/questions/10010000000000264
rename you script to main.py ,and save to sdcard /main.py.

[wangjianxin-canaan]

For other alternative, like Linux + RTOS is not yet available for V3.0 LP4 model @@
make CONF=k230_canmv_v3_defconfig can build 3.0 LP4 image;