Roll your own Pi Distro free of Microsoft 2021
PollenOS – Stretch based image: released 03142021 — Pi3B+
OpenCV face demo
Tensorflow Lite object detection demo
Note: we later realized Tensorflow uses contributor covenant and will not support/include future tensorflow demos
VNC Newer images include x2go for various reasons.
For this image, use a VNC client on your workstation/laptop etc.
user: pi passwd: raspberry
Remember to change the default passwd.
PollenOS – Bullseye based Image – experimental — Pi3B+
No Windows repos!
Yolov3 object detection – tiny demo – Test drive YOLOv3 tiny on Pi3B+
X2Go for remote desktop
Free Culture videos for running inference included
Imagenet working demo – Run predictions on included data images and more!
Ubuntu Groovy Image – Pi4
- Face Detection python demo – NCS2
- Darknet Yolov3 demo
In response to the inclurision of Microsoft repos and software in the most recent release of Raspberry Pi OS by the Raspberry Pi Foundation, we created a Raspberry Pi OS – Pollen with AI demos for Pi.
Debian Version: bullseye sid (experimental)- 3B+
imagenet – 3/20/2021
GUI – desktop environment LXDE All But AI completed – 2/25/2021
Ubuntu Version: 20.10 Groovy- 4B with 4Gb
GUI Desktop Environment – lubuntu
At this point in the journey, I will be building opencv from source.
Upon successful completion, I will:
- Test Darknet
- Test OpenVINO demos
- Clean files/add additional
Installing OpenCV 4.4.0 on Pi4 with 4Gb
- dphys-swapfile 4096 maxswap
- rpi-eeprom-update check… -a if needed
- config.txt gpu mem=128
- Depends – then to “dirty work (github)
- wget opencv
- wget opencv_contrib
- mv opencv-4.4.0 opencv_contrib-4.4.0 to opencv/
- mkdir build
- cmake ..
- make -j4
- finish up – make install ldconfig make clean apt update
- confirm opencv version – python3 cv2__version__