Raspberry Pi AI Microscope “for nuisance detection”

[magallonbuckinbulls]

Howdy y’all.

Since I'm currently always traveling I’m limited on what I can do. So i decided to build an AI Microscope for detecting nuisance in the water column.
For this I will be using:
- Raspberry Pi 5 w/Haiolo AI 26 tops module
- AmScope MU Series 10MP USB microscope camera
- (current desktop microscope) till I figure out how to connect to tank. Water would be on a loop with a pause for analysis. This will happen only if Im successful creating the LLM’s.
- LLM (need to train model to analyze) there’s some models out there I can start with and tweak as I go.

I have all the equipment except for the Hailo module. Its on backorder.

Would love to hear your input.

Note: this is not another human replacement gadget. It's more of a reference tool for comparison, and I love tech

​

 

[Dan_P]

Howdy y’all.

Since I'm currently always traveling I’m limited on what I can do. So i decided to build an AI Microscope for detecting nuisance in the water column.
For this I will be using:
- Raspberry Pi 5 w/Haiolo AI 26 tops module
- AmScope MU Series 10MP USB microscope camera
- (current desktop microscope) till I figure out how to connect to tank. Water would be on a loop with a pause for analysis. This will happen only if Im successful creating the LLM’s.
- LLM (need to train model to analyze) there’s some models out there I can start with and tweak as I go.

I have all the equipment except for the Hailo module. Its on backorder.

Would love to hear your input.

Note: this is not another human replacement gadget. It's more of a reference tool for comparison, and I love tech

​

Sounds like an engaging project. Some thoughts.

Micro algae are kinda scarce unless the water is very cloudy with the organism. The system you are building may see nothing most of the time.

Nuisance micro algae are in aquaria all the time. Detecting them in the water might be meaningless unless there is a diagnostic model that relates the number per gallon to a bloom.

The micro algae that bother aquarist are benthic, congregating on surfaces. The number in the water might not change during a bloom.

Since most blooms are benthic, observing surfaces might be more fruitful or diagnostic. Observing an illuminated surface that is continually bathed in aquarium water might be a way to obtain a sample of whats growing on the aquarium surfaces. The illuminated surface would have to be replaced or cleaned periodically. Another way to concentrate organisms is the skimmer. Continuously monitoring skimmate might up the organism count to be useful, though.both live snd dead organisms will be present. And then there is the difficulty of keeping skimmate yield steady so the number of organisms observed reflects water content and not variation in skimmate production.

Looking forwards to the working “observatory”.

 

[magallonbuckinbulls]

Thanks brother for your comment. Highly appreciated.

Yes I was thinking the same thing last night as I was walking my dog lol. Definitely need cloudy water or from a filter or surface as you mentioned.

Doing this more for fun than a gadget to replace monitoring. And to continue learning AI LLM’s and Ollama while on my travels.

I have a robot arm I'm also working on for cleaning glass, feeding, monitoring, ect.
I will share all once im done with design and generate stl files.
Will run on Arduino and Raspberry Pi.

Thank you

 

[magallonbuckinbulls]

 

[EnterName]

Interesting project!

Let me know if I can help somehow. I'm quite passionate about microscopy as you can see in my microscopy thread.

For my latest post, I used a plankton sieve to collect samples. This could be a way for you to collect specimens from flowing water for a while and then analyze the collected specimens. However, the things you find might look quite different from what you would find in substrate and filter samples.

You will also need to take into account how frequent you observe something. A single dinoflagellate or some diatoms are nothing to worry about and they will always be present in a healthy reef tank. However, once their numbers increase they can become a pest.

 

[Kooma]

Howdy y’all.

Since I'm currently always traveling I’m limited on what I can do. So i decided to build an AI Microscope for detecting nuisance in the water column.
For this I will be using:
- Raspberry Pi 5 w/Haiolo AI 26 tops module
- AmScope MU Series 10MP USB microscope camera
- (current desktop microscope) till I figure out how to connect to tank. Water would be on a loop with a pause for analysis. This will happen only if Im successful creating the LLM’s.
- LLM (need to train model to analyze) there’s some models out there I can start with and tweak as I go.

I have all the equipment except for the Hailo module. Its on backorder.

Would love to hear your input.

Note: this is not another human replacement gadget. It's more of a reference tool for comparison, and I love tech

​

I’ve got a few RPis sitting around, following along!