Add instructions and/or support for multiple printers

[JoeDaddy7105]

I have two P1Ses setup with this integration. I tried just duplicating the automation and setting up the entities for each respective printer however I don't think it is actually working properly. I had a failure detection this morning with both printers running and stopping the print from the mobile notification caused it to stop printing on both printers even though only one had spaghetti and the other was fine. It may help to either clarify if this can actually support multiple printers and if so, how to set it up for multiple printers.

[soundman666]

Thank you @nberktumer for your work on this integration; it’s incredibly helpful!

I also have two printers, the P1S and the A1 Mini, and it would be fantastic if the Bambu Lab P1 - Spaghetti Detection integration could support multiple instances with separate sets of entities for each printer. This would allow us to choose specific configurations for each printer within the script template, making it much more flexible for those of us with multiple devices.

If there’s any way to implement or support this functionality, I’d be thrilled to see it added. Thank you again for all your hard work, and please let me know if I can assist with testing or feedback!

[nobodyguy]

I don't want to hijack this integration, but my integration supports this usecase if you want to try it meanwhile.
https://github.com/nobodyguy/obico_ml_ha_integration
But it only works with my version of Obico ML server HA addon and the multithreading isn't properly tested right now (received P1S a few days ago).

[nberktumer]

Hi @nobodyguy, I've checked your project and performing all the major operations in HA integration is a good idea and as you said, it may allow supporting multiple printers. However, if I'm not mistaken, you are calculating the average confidence of a single detection. Meaning, the ML model can detect multiple failures in a single image and you are calculating the average of multiple failures in the same frame. So, if there is a false positive in a detection, it would report a failure.

What Obico does is (also the blueprint automation in this project) it aggregates the confidence with the historic data so that even if the ML model returns a false positive detection in a single frame, it doesn't report there is a failure. If the detections constantly report failure, then the aggregation converges to a failure and notifies the user.

If you wish to take a look at the Obico's aggregation function, here are the codes:
https://github.com/TheSpaghettiDetective/obico-server/blob/496605a62fcb790097c510c151994e9b2bf020c1/backend/lib/prediction.py

https://github.com/TheSpaghettiDetective/obico-server/blob/496605a62fcb790097c510c151994e9b2bf020c1/backend/app/models.py#L410

[nobodyguy]

Hello @nberktumer,
thank you for the reply. Yes, you are correct and porting predictions is the first thing on my TODO list (it's mentioned in the README), so currently it's user's responsibility to address it until it's implemented :)