Python Thermal Image Processing Server

Image processing server for ThermalGram(thermalgram.com). ThermalGram is an Android and web application which uses FLIR One camera to take and share thermal selfies with social media. This sever recognizes thermal temperature of the picture in the request, returns a response with the calculated temperature rating.

Main dependencies

Software

Python version 2.7.x or higher
Flask version 0.10.x or higher
Gunicorn 19.3.x or higher
Nginx 1.4.6 or higher
httplib2 0.8 or higher
numpy 1.8.x or higher
scipy 0.13.x or higher
scikit-learn 4.x or higher
OpenCV 2.4.8 or higher

Hardware

FLIRONE Thermal accessory for Android Smartphone

Connect

chmod 400 thermal-hack.pem
ssh -i thermal-hack.pem ubuntu@ec2-54-208-206-110.compute-1.amazonaws.com

Setting the basic environment (Ubuntu 14.04)

Creat and change to a new user

adduser new-user-name
sudo adduser new-user-name sudo
su - new-user-name

Install packages

sudo apt-get update
sudo apt-get install build-essential
sudo apt-get upgrade
sudo apt-get install git ufw libreadline6 libreadline6-dev
sudo apt-get install cmake libatlas-dev libatlas3gf-base build-dep

Configure the Firewall

sudo ufw enable
sudo ufw default deny
sudo ufw allow ssh
sudo ufw allow 123
sudo ufw allow 80/tcp
sudo ufw enable

Setting Python environment

Set up source folder and enable virtual environment

sudo mkdir -p /var/www/
sudo chown thermal /var/www/
cd /var/www
mkdir virtual_env & cd virtual_env
sudo apt-get install python-virtualenv
virtualenv . --system-site-packages

Run source bin/activate to enable the virtual environment.

python-setuptools python-pip

Install python

sudo apt-get install python-dev

Install OpenCV

In your virtual env:

sudo ln /dev/null /dev/raw1394
sudo apt-get install python-numpy libatlas-dev libatlas3gf-base
sudo apt-get install python-numpy python-scipy
sudo apt-get build-dep python-matplotlib
sudo apt-get install python-opencv
pip install --user --install-option="--prefix=" -U scikit-learn

Test if opencv is installed properly.

$ python
import cv2
cv2.__version__

You should see the proper version.

Clone git

cd /var/www
git clone https://github.com/SsureyMoon/ThermalGram-Image-server
mv ThermalGram-Image-server thermal_app
cd thermal_app
mkdir data/model
git update-index --assume-unchanged settings/config.py

Install more dependencies

pip install -r requirements.txt

Setup your own authorization token

In settings/config.py:

AUTH_TOKEN = "<replace this>"

Setup front end javascript(Optional)

In static/upload.js:

var publicDNS = "<repalce this to the public DNS of your server>";

Web server setting

We are going to use Nginx and Gunicorn to server our application.

Run Gunicorn server

Gnuicorn server only serve local requests. Nginx will work as a proxy server to pass remote request to local gunicorn server.

gunicorn -b 127.0.0.1:8080 -w 3 app:app --name thermal_app --timeout 20 --log-file logs.log --log-level debug &

Config Nginx

sudo apt-get install nginx
sudo service nginx start
sudo nano /etc/nginx/sites-available/thermal_app

Copy and paste following lines in /etc/nginx/sites-available/thermal_app:

server {
    listen   80;
    server_name ec2-54-208-206-110.compute-1.amazonaws.com;
    location / {
        proxy_pass http://127.0.0.1:8080;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }
    location /static {
        alias  /var/www/thermal_app/static/;
    }
}

Enable user configuration and disable default configuration:

sudo unlink /etc/nginx/sites-enabled/default
sudo ln -s /etc/nginx/sites-available/thermal_app /etc/nginx/sites-enabled/thermal_app

Run Nginx

sudo service nginx relaod
sudo service nginx restart

API endpoints

GET /:
- Response (JSON):
  - result: "success"
GET /image
- Response (HTML):
  - form for upload image file.
POST /image
- Form Data:
  - _auth_token: String
  - rate: Int
  - justimage: String
- Response(JSON):
  - image_file: original image file link, same as 'justimage'
  - rate: user rating (1 to 5)
  - result (Property):
    - how_other_user_say:3
    - images (Property):
      - orig:original image file url in this image server
      - base : Area with temperatures above threshold,
      - point1: Area with middle temperature
      - point2: Area with higher temperature

Test

Use _auth_token, which you set. Send post message in CMD to local server.

curl -X POST --form "_auth_token=<replace this>" --form "rate=4" --form "justimage=@/path/to/the/image/IMG_17.JPEG" http://127.0.0.1:8080/image/
curl -X POST --form "_auth_token=<replace this>" --form "rate=4" --form "justimage=http://thermalgram.com/julia.jpg" http://127.0.0.1:8080/image/

or post message in CMD to amazon server.

curl -X POST --form "_auth_token=<replace this>" --form "rate=4" --form "justimage=@/path/to/the/image/IMG_17.JPEG" http://ec2-54-208-206-110.compute-1.amazonaws.com/image/
curl -X POST --form "_auth_token=<replace this>" --form "rate=4" --form "justimage=http://thermalgram.com/julia.jpg" http://ec2-54-208-206-110.compute-1.amazonaws.com/image/

or Visit http://ec2-54-208-206-110.compute-1.amazonaws.com/image/