Motion detection API

This project has been developed to provide a small AI capability to motionEye security system and help to analyse if the motion came from a person moving in the field. In case of recognized person, it will automatically send and email with the attached picture.

Motivation

The main goal, on top of learning more about deep learning, was to add an local AI capability to the already great MotionEye project without having to pay for cloud API processing. This project uses the open source TensorFlow Lite as AI to recognize the motion. An email/alert will be send only if a person has been seen, not if a light turned on or a cat passed by.

Technologies and Packages used

The small server has been created as an API server that run the PI and start processing images by Path (located on the PI) or by URL.

NodeJs, Express, Typescript for the API server
Tensorflow Lite as AI capability to run the model locally and test the picture
nodemailer to send emails with pre-configured SMTP server
PM2 To start eh service in production ready and configure the service to restart then the pi restart
raspap To create a GUI to configure the wifi hotspot of the different camera (raspap.com)[https://raspap.com/]

Configure in a server

If you don't want to configure and run this project on a raspberry pi, you can use docker to deploy the app.

update the .env file with your SMTP variables
Run docker:

docker build -t me/motion-detection-api .
docker run -p 5000:5000 me/motion-detection-api

Configure in a Raspberry PI

NOTE: This project is still in progress, while the configuration of the pi with the motionEye script works great, the motion-detection-api service still has some performance issue. You need to use @tensorflow/tfjs instead of @tensorflow/tfjs-node since the node version does not install well in PI as of now. This result in an important performance drop.

Script configuration of MotionEye and motion-detection

Download raspbian light,
Use balenaEtcher to burn the image into the SD card,
Add ssh or ssh.txt file into the /boot folder before removing the SD card,
Put the SD card in the Pi and connect it to the wired network,
Connect with ssh: ssh pi@raspberrypi password: raspberry.

To help getting started, I created an install script raspbian-setup.sh. From a fresh rasbian install, it configures all you need for MotionEye to work with few more dependencies (Nodejs, C++ compiler). This script need to be run as sudo to work properly.

wget --no-check-certificate --content-disposition https://github.com/R0muald/motioneye-detection/archive/master.zip
unzip motion-detection-api-master.zip
mv motion-detection-api-master motion-detection
cd ./motion-detection
sudo chmod +x raspbian-setup.sh
sudo ./raspbian-setup.sh

restart your py:

sudo reboot

Update .env.local with your own MAIL configuration then run it:

cd ./motion-detection
yarn install
yarn build
pm2-runtime pm2-process.json
# boot strategy
pm2 startup
pm2 save

configure motionEye

From MotionEye web interface (http://<IP_PI>:8765), you can now configure the Motion Notification web hook with POST (form) and the URL to http://localhost:5000/notify/?path=<SNAPSHOT_URL>&email=<EMAIL_ADDRESS>.

Configure pi as a hotspot with raspap

Once Raspap installed, you will need to open the web interface and configure the hotspot:

Basic Tab:
- SSID: <YOUR_SSID_NAME> (store it somewhere)
- Wireless mode: 802.11n - 2.4GHz (to be compatible with the Raspberry Pi zero W)
Security Tab:
- type: WPA
- encryption: CCMP
- PSK: <YOUR_PSK_KEY> (store it somewhere)
Advanced Tab:
- Hide SSID in broadcast: Enabled

Configure Pi zero with motionEyeOS

Download the latest stable release img file (https://github.com/ccrisan/motioneyeos/releases)
Use Ether.io to burn a SD card with the image
Before putting the card in the RPI Zero,
1. If you want to access the pi with ssh, create an empty ssh.txt or ssh file at the root of the SD card (usually called /boot)
2. Create a file called wpa_supplicant.conf with:
```
update_config=1
ap_scan=1
ctrl_interface=/var/run/wpa_supplicant
network={
  scan_ssid=1
  ssid="<YOUR_SSID_NAME>"
  psk="<YOUR_PSK_KEY>"
  key_mgmt=WPA-PSK
}
```

FrancoisDF/motion-detection-api