
Hello guys, just an update of my [last post](/r/belgium/comments/tk2gf0/waterlink_smart_meters/) about the smart meters installed by Waterlink.
Well, Waterlink answered my message which is already more than I expected, so that’s something I guess. But they just answered that “for now it’s not possible to link the digital meter data to other systems via APIs”.
So that’s for accessing the API, and they ignored my main question which was about the meter’s AES public key for receiving the data myself. I’m going to send them an answer about this but of course I have little hope of anything.
So as an alternative, I’ve set up something unsatisfactory that works more or less nonetheless. It’s an ESP32-CAM affixed to the bottom of an upside down white plastic drinking cup, with a cutout to let just the camera through, which conveniently has the right diameter and height to fit the water meter display at the right distance for the camera.
I get [this kind of picture](https://seos.fr/blog/images/compteur-eau.jpg) which is a bit shitty because of the low-contrast display and non ideal placement of the flash, but is actually good enough to OCR. The [AI-on-the-edge](https://github.com/jomjol/AI-on-the-edge-device) project didn’t work well for me because it takes ages to process one photo and I want better precision than that, so I rigged up a script that uses imagemagick to rotate, crop and threshold the image to [something like that](https://seos.fr/blog/images/compteur-eau-nb.jpg) and then [ssocr](https://www.unix-ag.uni-kl.de/~auerswal/ssocr/) ([github](https://github.com/auerswal/ssocr)) does the job of actually [recognizing the numbers](https://seos.fr/blog/images/compteur-eau-ocr.png).
It works well most of the time, and I take a picture every 5 seconds so the precision is quite good. In typical use, during a shower for example, the meter updates its display for every 2 liters used. There might be a little more info on [my blog](https://seos.fr/blog/20220331.kamstrup.html) but it’s in French and doesn’t give a lot more details regarding the actual solution, so you can probably skip it.
I’d still prefer a solution that is immune to the inevitable errors of the OCR process, but for now this will do OK. That’s all, folks.
2 comments
Don’t they have a system similar to that of fluvius where you can get the real time data from their website for gas/electricity consumption? If they have the data I don’t see what’s the issue?
I have no experience/knowledge of this device, so maybe you can clarify;
-Is it a part of your own netwerk (wifi/cable), so this device can upload metrics to waterlink?
-Does this device have any kind of console/webgui or anything?