qengineering / rock-5-ubuntu-22-image

Rock Pi 5 image with Ubuntu 22, OpenCV, deep learning frameworks and NPU drivers

Rock 5 with Ubuntu 22.04 OS image

---

Installation.

• Get a 16 GB (minimal) SD card holding the image.
• Download the Rock5_Ubuntu22.img.xz image (3.6 GByte) from our Sync site.
• Flash the image on the SD card with the Imager or balenaEtcher.
• Insert the SD card in your Rock 5 and enjoy.
• Username: 5ubuntu22
• no password

---

Model performance benchmark(FPS)

All models, with C++ examples, can be found on the SD image.

demo	model_name	inputs_shape	dtype	Rock 5
yolov5	yolov5s_relu	[1, 3, 640, 640]	INT8	50.0
	yolov5n	[1, 3, 640, 640]	INT8	58.8
	yolov5s	[1, 3, 640, 640]	INT8	37.7
	yolov5m	[1, 3, 640, 640]	INT8	16.2
yolov6	yolov6n	[1, 3, 640, 640]	INT8	63.0
	yolov6s	[1, 3, 640, 640]	INT8	29.5
	yolov6m	[1, 3, 640, 640]	INT8	15.4
yolov7	yolov7-tiny	[1, 3, 640, 640]	INT8	53.4
	yolov7	[1, 3, 640, 640]	INT8	9.4
yolov8	yolov8n	[1, 3, 640, 640]	INT8	53.1
	yolov8s	[1, 3, 640, 640]	INT8	28.5
	yolov8m	[1, 3, 640, 640]	INT8	12.1
yolox	yolox_s	[1, 3, 640, 640]	INT8	30.0
	yolox_m	[1, 3, 640, 640]	INT8	12.9
ppyoloe	ppyoloe_s	[1, 3, 640, 640]	INT8	28.8
	ppyoloe_m	[1, 3, 640, 640]	INT8	13.1
yolov5_seg	yolov5n-seg	[1, 3, 640, 640]	INT8	9.4
	yolov5s-seg	[1, 3, 640, 640]	INT8	7.8
	yolov5m-seg	[1, 3, 640, 640]	INT8	6.1
yolov8_seg	yolov8n-seg	[1, 3, 640, 640]	INT8	8.9
	yolov8s-seg	[1, 3, 640, 640]	INT8	7.3
	yolov8m-seg	[1, 3, 640, 640]	INT8	4.5
ppseg	ppseg_lite_1024x512	[1, 3, 512, 512]	INT8	27.5
RetinaFace	RetinaFace_mobile320	[1, 3, 320, 320]	INT8	243.6
	RetinaFace_resnet50_320	[1, 3, 320, 320]	INT8	43.4
PPOCR-Det	ppocrv4_det	[1, 3, 480, 480]	INT8	31.5
PPOCR-Rec	ppocrv4_rec	[1, 3, 48, 320]	FP16	35.7

• Due to the pixel-wise filling and drawing, segmentation models are relatively slow

---

Tips.

• If you need extra space delete the opencv and the opencv_contrib folder from the SD card. They are no longer needed since all libraries are stored in the /usr/ directory.
• Use a tool like GParted sudo apt-get install gparted to expand the image to larger SD cards. We recommend a minimum of 64 GB. Deep learning requires a lot of space.
  

---

Pre-installed frameworks.

• OpenCV 4.8.0
• ncnn 20240410
• NPU rknpu2 1.5.2
• NPU model zoo 2.0.0
• NPU model zoo models 2.0.0

---

Thanks.

A more than special thanks to Stuart Naylor, who, ever so kindly, provided us with the Rock Pi 5 for free.
And Joshua Riek for all the hard work on the Ubuntu OS.

---

rock-5-ubuntu-22-image's Issues

Question about FPS data

Hii,

As usual thank you so much for what you are doing for the community!

I wanted to know if the FPS values shown in the table are for model inference on NPU or with ncnn?

I have run custom yolov5s models on the orangepi5 using npu but it only gives me upto 25FPS. Since both are RK3588, and theres very little performance difference between RK3588 and 3588s, I want to know what can be done to get my FPS to 37 and above.

Thanks in advance!

4k radxa csi camera drivers preinstalled ?

hi!
thanks for image sharing .
but i`m interesting if native csi camera for radxa is working OK on this ubuntu - https://wiki.radxa.com/Rock5/accessories/radxa-camera-4k

i know there`s some problem with camera driver on ubuntu it shows green colors - https://www.youtube.com/watch?v=wwiR1BqJFoM
and there is no official ubuntu release from radxa (only cli image without desktop and drivers).