v-k-k / ai_smart_pi_doorbell

Smart Doorbell - Face-Recognition and notification by MQTT

AI Smart Doorbell

Bell was created with the intention of using minimal local physical resources in this way every notification process is by MQTT

Diagram:

Features

• Face Recgonition via azure
• Notification by MQTT
• Action Callback by MQTT
  • Live
  • Snapshot
  • Playsound
  • Add new person to Face Recognition
• Call and receive Voipcall

Hardware

In my case I used raspberry pi zero, but could use any other raspberry or orangepi PC (in the case of orangepi pc do not need to buy microphone and usb sound)

Hardware	Notes
Raspberry pi zero	You can buy another version
Power suply	Power supply you need to provide 5v for raspberry and 5v for speaker (in my case I put 12v converter with two outputs of 5v)
usb sound card (speakers and microphone)	Before buying any of them check if is compatible with raspberry, in the aliexpress there is a usb sound card 7.1 for 1 euro
touch button	It is advisable touch button becouse of rain, this being under the acrylic avoiding water leakage.
Raspcam Infrared	There are several camera to raspberry with various angles, should buy what suits for your front door.
Another machine	You need another machine on your internal network with mqtt server, http server with HTTP Basic authentication configure, node-red or homeassistant to send notifications wherever you want

Installation

Some python requirements or libraries may be missing if you find any steps failed please tell me.

First (External HTTP server)

Copy the index.html file to the http folder of the external server and in turn fill with your data. Note: The server must have HTTP Basic authentication

Second (Configure Raspberry)

Install all dependencies

sudo apt-get update
sudo apt-get upgrade
sudo apt-get install python-gpiozero
sudo apt-get install python-pip
sudo apt-get install sox
sudo apt-get install sox libsox-fmt-all
sudo apt-get install ffmpeg
sudo apt-get install subversion
sudo apt-get install -y gpac
sudo apt-get install linphone
sudo apt-get install lighttpd

Run the following command

sudo chmod 777 /var/www/html

Edit file lighttpd.conf and change server.stat-cache-engine to "disable"

sudo nano /etc/lighttpd/ lighttpd.conf 

Edit file "/etc/network/interfaces" to ensure wifi connection start in boot. Something like this:

auto lo
iface lo inet loopback
iface eth0 inet dhcp
auto wlan0
allow-hotplug wlan0
 
iface wlan0 inet static
    address 192.168.1.120
    netmask 255.255.255.0
    network 192.168.1.254
    broadcast 192.168.1.255
    gateway 192.168.1.254
    wpa-ssid "SIID_WIFI"
    wpa-psk "Password_wifi"

Run Github command in the user pi folder (/ home/pi)

git clone https://github.com/Hurleyking/AI_Smart_PI_Doorbell

To ensure WiFi is always on top in case of failure, copy the following file to the directory indicated below:

cp checkwifi.sh  /usr/local/bin/checkwifi.sh

Check alsa if this is using correct device (in my case usb1)

alsamixer

Edit the fixed variables in the files auto_healing.py and doorbell_pi_start.py:

http_server = 'http_server_adress_and_port_file'
http_password = 'username'
http_user = 'password'

Notice: To use face recognition it is necessary create groupId new, for do that you need discomment the line 348 from file doorbell_pi_start.py "#person_group_create_v2()" to "person_group_create_v2()" after run script you need comment again, otherwise try to always create new groupid whenever script starts

Add follow lines in crontab(sudo crontab -e)

*/5 * * * * /usr/bin/sudo -H /usr/local/bin/checkwifi.sh >> /dev/null 2>&1
*/10 * * * * python /home/pi/AI_Smart_PI_Doorbell/doorbell_pi/auto_healing.py

Ready to use.

MQTT instructions:

To get notification, subscribe to doorbell/ring

To get status from doorbell, subscribe to Doorbell/Temperature and Doorbell/rssi .

Publish doorbell/live:

Payload with the String live_30sec - Action: Record 30 seconds of live broadcast and save mp4 to folder /var/www/html/ live_30sec.mp4

Payload with the String live_30sec_alexa - Action: Record 30 seconds of live broadcast and save mp4 to folder /var/www/html/ live_30sec.mp4 (for alexa node-red)

Payload with the String photo - Action: takes a photo realtime and saves it in the folder/var/www/html/last_ring.jpg

Payload with the String {"new_person":"True","name":"marco"} - Action: add new person for facial recognition based on the last photo taken (/var/www/html/last_ring.jpg)

Todos

• Add condition to created groupid automatic
• tell me more function

Free Software, Hell Yeah!