Skip to content

Latest commit

 

History

History
145 lines (96 loc) · 5.81 KB

HOWTO.md

File metadata and controls

145 lines (96 loc) · 5.81 KB
Raw

How to turn your analog water meter into a Home Assistant sensor

Overview

The system contains of three components that communicate with HTTP:

  • Camera
  • watermeter
  • Home Assistant

demo image

Taking a photo of your water meter

You can use any camera you like as long as it shoots color images. Night vision cameras do not provide good results, as it's close to impossible to identify the analog gauges with a greyscale image. Therefore I use a standard camera and a white LED for illumination when taking photos.

The following example is using

  • A Raspberry Pi Zero W running Raspberry OS
  • A cheap Raspberry Pi camera
  • A bright white 3.3V LED connected to GPIO 17
  • Imagemagick for photo post processing. Install via sudo apt install imagemagick

In my experience, the worse the image quality, the easier it is for the OCR to read the digits. To see an example, how bad the quality can be, take a look at the demo image. Sometimes when the meter is fogged, the quality is even worse, but the results are still accurate.

From cron, I call watermeter.sh minutely. The script takes the photo and does some image enhancement using Imagemagick:

#!/bin/bash

python watermeter.py
convert -crop 906x906+759+1089 /run/shm/watermeter_last.jpg /run/shm/watermeter_crop.jpg # you need to adjust this to your situation
convert -contrast -equalize /run/shm/watermeter_crop.jpg /run/shm/watermeter.jpg # play around with different filters to find the best results

Aim of the image enhancement is to make the digits look as flat as possible and the gauges to look as red as possible. Images are stored in /run/shm to not wear the SD card.

watermeter.py is taking care of driving the LED and the camera:

from gpiozero import LED
from time import sleep
from picamera import PiCamera

led = LED(17) # LED is connected to GPIO 17 (3,3 V on a Zero)
    
camera = PiCamera()
camera.resolution = (2592, 1944)
camera.brightness = 60
# Turn on LED
led.on() 
# Turn on Camera and allow to adjust to brightness
camera.start_preview()
sleep(5)
# Take an image. I put in in /run/shm to not wear the SD card
camera.capture('/run/shm/watermeter_last.jpg')
camera.stop_preview()
led.off()

Serving the final meter image

I server the image from /run/shm using NGINX with disabled access logging to reduce SD card wearing.

  • Install nginx via sudo apt install ngnix
  • In /etc/nginx/nginx.conf change access_log /var/log/nginx/access.log; to access_log /dev/null;
  • In /etc/nginx/sites-enabled/default change root /var/www/html to root /run/shm;
  • Restart nginx via sudo systemctl restart nginx.service

Processing your photo to data using watermeter

General Considerations

I recommend to run it on a x86_64 machine, but a Raspberry Pi 3 should be sufficient. Lower powered devices like a Raspberry Pi Zero may also work, but I've never tested that.

Running watermeter with Docker Compose

The following example requires a running docker-compose setup. On Debian based systems install via sudo apt install docker-compose docker.io. I recommend to enable docker service to start on boot with sudo systemctl enable docker. You may also want to add your user to the docker group: sudo addgroup YOURUSER docker.

The docker-compose.yaml for running watermeter looks like this:

version: "3.5"
services:
  watermeter:
    image: nohn/watermeter:latest
    container_name: watermeter
    volumes:
      - ./watermeter/config:/usr/src/watermeter/src/config
    restart: unless-stopped
    ports:
      - "3000:3000"

After creating the config directory using mkdir -p watermeter/config you can start watermeter with docker-compose pull && docker-compose up -d.

Configuration

After watermeter has started, you can access the configuration UI at http://ip.of.docker.host:3000/configure.php

Configuration GUI Screenshot

Replace "Source Image" with http://ip.of.camera.pi/watermeter.jpg and hit the "preview" button. Now, step by step enter the coordinates of your digits and gauges and verify with "preview". Once you are done, click the "save" button to save your configuration.

Verify the results by accessing http://ip.of.docker.host:3000/?debug.

If the digits and gauges are placed correctly, but the results are wrong, please try different options for enhancing the image (see above).

Once you are happy with the results, you can access the plain number on http://ip.of.docker.host:3000/.

Adding a sensor to Home Assistant

You need a running Home Assistant instance. Please referer to https://www.home-assistant.io/installation/ for installation instructions. I strongly recommend not to run it on a Raspberry Pi and definitly not on a SD card. Home Assistant is constantly writing to disk and will wear any SD card within a few months. I recommend using some cheap and low power x86_64 machine with a SSD for that.

Once you have set up your Home Assistant, please add this to your configuration.yaml:

sensor:
  - platform: rest
    name: Water
    resource: "http://ip.or.hostname.of.watermeter:3000/"
    scan_interval: 60
    unit_of_measurement: 'm³'

And restart Home Assistant.

After that you should have a new sensor call "Water".

Home Assistant Dashboard

How to contribute

You can contribute to this project by:

  • Opening an Issue if you found a bug or wish to propose a new feature
  • Placing a Pull Request with bugfixes, new features etc.

You like this?

Consider a gift.

License

analogmeterreader is released under the GNU Affero General Public License.