This is an overview of different project part to digitalize an analog water meter and use it for a house automatization system to controll the water consumption.
UPDATE: new Version for ESP32-CAM available on Thingiverse - much easear and cheaper due to integrated camera and illumination https://www.thingiverse.com/thing:3860911
The system consists of a hardware fixture and and successiv software code to transform this finally to a measurement value for a database input.
To get an running system one needs 3 components:
| Component | Purpose | Link |
|---|---|---|
| 1. Mechanical Fixture | Hardware components to fix a camera and illumination on a water meter | Version for ESP32-CAM Version for ESP8266 + ArduCAM |
| 2. Taking pictures | ESP8266 based html server to provide pictures from the camera | https://github.com/jomjol/water-meter-picture-provider |
| 3. Server for data processing and evaluation | Image processing (Cutting, OCR, ...) to extract the needed features | https://github.com/jomjol/water-meter-system-complete |
- Server in Docker-Container https://cloud.docker.com/u/jomjol/repository/docker/jomjol/wasserzaehler
- GitHub-Details on Server https://github.com/jomjol/docker-wasserzaehler
Extended user discussion can be found in the following forum: https://forum.iobroker.net/topic/23577/wasserzähler-selfmade
| Component | Usage | Link |
|---|---|---|
| Alignment and decomposition | Image processing with OpenCV library to extract the needed features | https://github.com/jomjol/water-meter-image-cut |
| OCR for full digits | Usage for OCR to extract the m³ part of the water meter | https://github.com/jomjol/neural-network-digital-counter-readout |
| Analog to Digital Readout | Usage of a Neural Network to digitalize analog pointers | https://github.com/jomjol/neural-network-analog-needle-readout |