Now that I have my Pi5 8GB it is time to try AI, CV, LMM etc.
While I have put this in the "Advanced users " section, I found it very easy, so should it be in the Basic user section?
Easy if you have a Pi5 with 8GB?
I would not like not try it on any smaller Pi
LLM - Large Language Models
A search for "Ollama Raspberry Pi", finds two starter links
https://fleetstack.io/blog/running-open ... ith-ollama
https://itsfoss.com/raspberry-pi-ollama-ai-setup/
I suggest you have a SSD as these models are Large...
Just some with more to collect![:lol:]( "Laughing")
Ollama will find and download the LLM the first time they are run.
A fast internet is handy and these are big downloads, 4GB+ for some.
A search on YT will find those.
CV - Computer Vision
There are many ways to get computer vision tools installed on Pi5 but it not as easy as it used to be with the Bookworm and Python/Pip changes.
I found a simpler way, install Processing first.
https://processing.org/download
Java on a Pi5 is now fast enough to be useful or less painful.![;)]( "Wink")
Once you have Processing installed go to Tools - Manage Tools and look at a bunch of handy CV libraries.
Weka, Deepvison, Creative Machines, Image processing algorithims, OpenCV (a very easy install unlike the "normal" install.)....
While some AI model/tools will download and install when the examples are run some do have issues like Tensorflow, MXNet.....
So while Face and Pose detection examples in OpenCV, Deepvision work, there are issues with Creative Machines examples.
Grab the webcam apps too, they work with USB cams, many years ago I got some Sony Eyecams for $2 each.
Hmm, does the Pi5 have enough power to run 4 Eyecams?
There are a bunch of other fun libs to install and example to play with![:D]( "Very Happy")
Hmm, can Processing make Android apps on the Pi5?
Anyway, this stuff is now easy to get started with a Pi5 with 8GB.
Have fun.
While I have put this in the "Advanced users " section, I found it very easy, so should it be in the Basic user section?
Easy if you have a Pi5 with 8GB?
I would not like not try it on any smaller Pi
LLM - Large Language Models
A search for "Ollama Raspberry Pi", finds two starter links
https://fleetstack.io/blog/running-open ... ith-ollama
https://itsfoss.com/raspberry-pi-ollama-ai-setup/
I suggest you have a SSD as these models are Large...
Just some with more to collect
Ollama will find and download the LLM the first time they are run.
A fast internet is handy and these are big downloads, 4GB+ for some.
- ollama list
NAME ID SIZE MODIFIED
codellama:latest8fdf8f752f6e3.8 GB2 hours ago
llama2:latest 78e26419b4463.8 GB13 hours ago
llava:latest 8dd30f6b0cb14.7 GB17 hours ago
mistral:latest 61e88e8845074.1 GB13 hours ago
phi:latest e2fd6321a5fe1.6 GB16 hours ago
tinyllama:latest2644915ede35637 MB17 hours ago
A search on YT will find those.
CV - Computer Vision
There are many ways to get computer vision tools installed on Pi5 but it not as easy as it used to be with the Bookworm and Python/Pip changes.
I found a simpler way, install Processing first.
https://processing.org/download
Java on a Pi5 is now fast enough to be useful or less painful.
Once you have Processing installed go to Tools - Manage Tools and look at a bunch of handy CV libraries.
Weka, Deepvison, Creative Machines, Image processing algorithims, OpenCV (a very easy install unlike the "normal" install.)....
While some AI model/tools will download and install when the examples are run some do have issues like Tensorflow, MXNet.....
So while Face and Pose detection examples in OpenCV, Deepvision work, there are issues with Creative Machines examples.
Grab the webcam apps too, they work with USB cams, many years ago I got some Sony Eyecams for $2 each.
Hmm, does the Pi5 have enough power to run 4 Eyecams?
There are a bunch of other fun libs to install and example to play with
Hmm, can Processing make Android apps on the Pi5?
Anyway, this stuff is now easy to get started with a Pi5 with 8GB.
Have fun.
Statistics: Posted by Gavinmc42 — Thu Feb 22, 2024 11:31 am