Raspberry Pi app using a camera and PIR motion sensor, written in Node.js using Johnny-Five and KittyDar for with cat facial detection.

I will write up the step-by-step tutorial (hopefully) soon! But unti then, here is the instruction how to run this code locally with your own Raspberry Pi.

• Raspberry Pi 2 (with Raspbian. Also with WiFi adapter)
• 5MP Camera Board Module ([buy])
• Pyroelectric Infrared (PIR) motion sensor ([buy])

If you are a Raspberry Pi newbie, I recommend to buy your first Pi from [CanaKit].

• 1 red wire: PIR-VCC to Pi's 3V
• 1 black wire: PIR-GND to Pi's ground
• 1 whatever color wire: PIR-OUT to Pi's Pin 7 (GPIO 4)

$ sudo apt-get update

then

$ sudo apt-get upgrade

$ wget http://node-arm.herokuapp.com/node_latest_armhf.deb

once downloaded, install

$ sudo dpkg -i node_latest_armhf.deb

Check if node is successfully installed

$ node -v

Go to Pi Software Config Tool to enable camera

$ sudo raspi-config

Test if your camera is working by try typing this command on terminal:

$ raspistill -o photo.jpg

I would like to say, $ npm install to install all the dependencies, and voilà! but it is not!

for cat facial detection, I am using kittydar, which dependencies including node-canvas, which requires Cairo.

So let's get Cairo on your Raspbian first.

$ sudo apt-get install libcairo2-dev libjpeg8-dev libpango1.0-dev libgif-dev build-essential g++

See more info on how to install Cairo for Node Canvas, see this Installation Ubuntu and other Debian based systems

If you download and use the whole node_modules contents off this repo, skip the step 2, and proceed to step 3. Otherwise, go to the next step to fresh-install the next several modules.

Once your environment is set up, in this RPi-KittyCam dir, install node dependency modules.

Ideally install from npm install kittydar —save

However, node-canvas 1.0.1 (the version specified in package.json for KittyDar) failed to build with the current Node.js (v0.12.6).

So what I did was download the zip from github repo into node_modules, alter the package.json, where canvas: ~1.0.1 to ^1.0.1 so that the latest canvas is installed as I npm install from the kittydar directory.

Get the zip from my forked repo.

Note: I am sending a pull request (harthur/kittydar#27)

$ npm install johnny-five

For realtime live-updating the web interface, use PubNub.

$ npm install pubnub

You need to sign up and get you own publish and subscribe keys!

For storing photos, use Cloudinary.

$ npm install cloudinary

You need to sign up and get you own API keys!

Create a config.js in the root dir of the app. The file should include your API keys:

module.exports = {

  cloudinary: {
    cloud_name: 'your_name',
    api_key: 'your_API_key',
    api_secret: 'your_API_secret',
  },

  pubnub: {
    subscribe_key: 'your_sub_key',
    publish_key: 'your_pub_key'
  }

};

You must run with sudo:

$ sudo node kittyCam.js

The camera will take a photo when a motion is detected by the PIR sensor. Then the child_process runs to detect if there is any cats in the photo. When there are any cat, it sends the photo to Cloudinary.

Analized photos are deleted from the filesystem to clear up Pi.

• Get the web interface source code from gh-pages branch.
• Run the index.html on browser

• No photos during mealtime. When a cat is eating food (head-down position), the facial detection doesn't recognize the cat at all.
• Raspistll continuously takes a bunch of photos when I set t = 0 (and crashes Pi while so many child process is running) so I have set t = 1, which causes delay. It seems to take only integer. Cats are too fast to wait for a second.