Code to read weight measurements from Xiaomi Body Scales.
Please note there was a breaking change in 0.1.8. The MQTT message json attributes are now in lower snake_case to be compliant with Home-Assistant Attributes.
This means Home-Assistant sensor configuration needs to be adjusted.
For example
value_template: "{{ value_json['Weight'] }}"
Needs to be replaced with
value_template: "{{ value_json['weight'] }}"
(note the lowercase w in weight)
| Name | Model | Picture |
|---|---|---|
| Mi Smart Scale 2 | XMTZCO1HM, XMTZC04HM | |
| Mi Body Composition Scale | XMTZC02HM | |
| Mi Body Composition Scale 2 | XMTZC05HM |
If using Home Assistant (formerly known as hass.io), try instead the Xiaomi Mi Scale Add-On for Home Assistant based on this repository.

docker-compose.yml (see below) and edit the environment to suit your configuration...docker-compose up -dversion: '3'
services:
mi-scale:
image: lolouk44/xiaomi-mi-scale:latest
container_name: mi-scale
restart: always
network_mode: host
privileged: true
environment:
- HCI_DEV=hci0 # Bluetooth hci device to use. Defaults to hci0
- MISCALE_MAC=00:00:00:00:00:00 # Mac address of your scale
- MQTT_HOST=127.0.0.1 # MQTT Server (defaults to 127.0.0.1)
- MQTT_PREFIX=miscale # MQTT Topic Prefix. Defaults to miscale
- MQTT_USERNAME= # Username for MQTT server (comment out if not required)
- MQTT_PASSWORD= # Password for MQTT (comment out if not required)
- MQTT_PORT= # Defaults to 1883
- TIME_INTERVAL=30 # Time in sec between each query to the scale, to allow other applications to use the Bluetooth module. Defaults to 30
- MQTT_DISCOVERY=true # Home Assistant Discovery (true/false), defaults to true
- MQTT_DISCOVERY_PREFIX= # Home Assistant Discovery Prefix, defaults to homeassistant
# Auto-gender selection/config -- This is used to create the calculations such as BMI, Water/Bone Mass etc...
# Up to 3 users possible as long as weights do not overlap!
# Here is the logic used to assign a measured weight to a user:
# if [measured value in kg] is greater than USER1_GT, assign it to USER1
# else if [measured value in kg] is less than USER2_LT, assign it to USER2
# else assign it to USER3 (e.g. USER2_LT < [measured value in kg] < USER1_GT)
- USER1_GT=70 # If the weight (in kg) is greater than this number, we'll assume that we're weighing User #1
- USER1_SEX=male # male / female
- USER1_NAME=Jo # Name of the user
- USER1_HEIGHT=175 # Height (in cm) of the user
- USER1_DOB=1990-01-01 # DOB (in yyyy-mm-dd format)
- USER2_LT=35 # If the weight (in kg) is less than this number, we'll assume that we're weighing User #2
- USER2_SEX=female # male / female
- USER2_NAME=Serena # Name of the user
- USER2_HEIGHT=95 # Height (in cm) of the user
- USER2_DOB=1990-01-01 # DOB (in yyyy-mm-dd format)
- USER3_SEX=female # male / female
- USER3_NAME=Missy # Name of the user
- USER3_HEIGHT=150 # Height (in cm) of the user
- USER3_DOB=1990-01-01 # DOB (in yyyy-mm-dd format)
Note: Python 3.6 or higher is required to run the script manually
wrapper.sh and configure your environment variables to suit your setup.@reboot bash /path/to/wrapper.sh
NOTE: Although once started the script runs continuously, it may take a few seconds for the data to be retrieved, computed and sent via mqtt.
Under the sensor block, enter as many blocks as users configured in your environment variables:
- platform: mqtt
name: "Example Name Weight"
state_topic: "miscale/USER_NAME/weight"
value_template: "{{ value_json['weight'] }}"
unit_of_measurement: "kg"
json_attributes_topic: "miscale/USER_NAME/weight"
icon: mdi:scale-bathroom
- platform: mqtt
name: "Example Name BMI"
state_topic: "miscale/USER_NAME/weight"
value_template: "{{ value_json['bmi'] }}"
icon: mdi:human-pregnant
unit_of_measurement: "kg/m2"


Thanks to @syssi (https://gist.github.com/syssi/4108a54877406dc231d95514e538bde9) and @prototux (https://github.com/wiecosystem/Bluetooth) for their initial code
Special thanks to @ned-kelly for his help turning a "simple" python script into a fully fledged docker container
Thanks to @bpaulin for his PRs and collaboration
Content type
Image
Digest
Size
190.4 MB
Last updated
almost 6 years ago
docker pull aiir42/xiaomi-mi-scale