OpenCV python program with a simple webpage to track weight readings from a scale.

This project:

• Takes an image from a webcam when scale begins weighing
• Process the image to both reduce noise and crop to the specific section of the display
• Do OCR on the image
• Record data
• Graph data on a web page

To start, you will need a copy of this project. Open up your terminal and then you will be set to start! First, you will need to install git if you don't have it yet: sudo apt-get install git. Then type git clone https://github.com/amcolash/ScaleIoT.git or git clone git@github.com:amcolash/ScaleIoT.git, depending on your preferences for cloning repos. If you don't know, choose the first option.

• Install opencv (with python bindings and NumPy for python 2.x) sudo apt-get install python-opencv python-numpy
• Install ssocr (ocr of seven segment displays)

sudo apt-get install libx11-dev libimlib2-dev make
git clone https://github.com/auerswal/ssocr.git
cd ssocr/
make
sudo make install

I had some problems compiling ssocr, after the make command I then ran cc ssocr.o imgproc.o help.o -L/usr/lib/arm-linux-gnueabihf -lImlib2 -lm -o ssocr followed by another make and finally sudo make install. Yikes, compiler linking problems (and I am pretty rusty with Makefiles).

To install this ScaleIoT as a service, edit the scaleiot script in the root of this repository with vim scaleiot or nano scaleiot. Then just run ./install.sh from the root directory of the project. (If you do not have upstart, you may need to add to rc.local or something similar)

In this project I am using nginx because it is lightweight and less overhead for the pi. The setup is also really simple!

First, install with sudo apt-get install nginx. Then, to change the deafult location for the server, do sudo vim /etc/nginx/sites-available/default (substitute for your favorite editor). On the line root /var/www/html, just edit the path to the root directory of the project.

Finally, you will need to install dependencies for the web page. I used bower in this case. You might need to install that and nodejs, check out AdaFruit's wonderful tutorial and simple setup for node (on pi 1 + 2, pi 3 just do sudo apt-get install node-js).

npm install -g bower
bower install

Finally, reboot: sudo reboot.

• OpenCV for Python - Open source computer vision library
• NumPy - Python computation and array library
• ssocr - Seven segment display digit detection
• HighStock - Javascript library for graphing data
• highcharts-regression - Regression lines for Highcharts / HighStock
• Bootstrap - CSS and Javascript library for layouts
• Bower - Front-End web dependency management