S-JZ / CFEngage

A violence detection module that can inform authorities about the details of individuals present at the site of the incident using face recognition.

Crime Free

A violence detection module that can inform authorities about the details of individuals present at the location during the incident using face recognition.

Tech Stack Used:

• Django
• MySQL
• Tensorflow
• Python 3.7
• OpenCV
• face_recognition
• HTML/CSS/JS
• pywhatkit
• yagmail
• git

How to set up the application?

• Clone the GitHub repository

$ git clone https://github.com/S-JZ/CFEngage.git

• Change directory to crime_free

$ cd CFEngage/crime_free

• Set up the virtual environment

$ virtualenv venv
$ source venv/Scripts/activate

• Install dependencies from requirements.txt

$ pip install -r requirements.txt

• Set up the Django Server

$ python manage.py makemigrations
$ python manage.py migrate
$ python manage.py runserver

• Open your web-browser and type in localhost:8000/.

Note: Before testing the application

• You might want to create a super user for the admin page in Django to add people in the database manually.

$ python manage.py createsuperuser

• Enter your desired username, email id and password for the same.

$ python manage.py runserver

• Open your web-browser and type in localhost:8000/admin/. You will be asked to login to the dashboard. Login with the above credentials.
• Now add in the information about the entries in the faces_db folder in the Persons model from the interface manually. By default, there are two images in the faces_db folder.
• Create the dummy Aadhar number, Name and Address of these two persons using the interface.
• You are now set to test the app!

Environment Variables:

The code uses sensitive information like passwords and secret keys which have been stored in the environment variable for security concerns. These can be re-produced on your system by defining the following in a .env file in the crime_free subdirectory:

DJANGO_SECRET_KEY 
DEBUG              #set to True
DB_NAME            # name of the database (for e.g. dummy_faces)
DB_USER            # database username
DB_PSW             # database password
DATABASE_URL       # localhost
DEVELOPMENT_MODE   #set to True
email    #own gmail id for sending mail to authorities
psw      #password of gmail account
polemail #police authorities email id (create a dummy one or supply own)
phone    #police whatsapp number 

Note:

• Ensure that you are logged in to the web whatsapp on the browser before uploading the video for processing.
• The mailing feature is by default not activated, if you wish to use it, make sure to enable less secure apps settings in your gmail account and set mail=True in the message function call in smart_patrol/FaceRecognition.py file.
• It is recommended to use <=360p video files for better performance. You can use the fight2.mp4 file for testing, added in the parent directory of the repository.

Demo Video Link:

The demo video is at: Microsoft Engage 2022 Demo Video

Design Documentation

The following document contains the data flow diagram and a simple tech stack diagram to showcase the architecture of the application.

• Design Document

Weekly Reports: Sprints

The file contains the weekly targets and milestones that I worked on throughout the month of May 2022 for the project. These weekly sprints include researching about different tech-stacks, implementing code, reviewing design and testing the application in each iteration.

• Weekly Targets Report

The learning sprint involved getting familiar with new tech stacks, and implementing a test driven development for the project.

Future Scope:

1. Reviewing the design to enable faster processing using multiple threads and process synchronization -> parallel processing.
2. Automating database connection with face database.

Dataset used for Violence Detection:

The dataset used for model creation and preprocessing the video was RWF2000-Video-Database for Violence Detection which contains 2000 clips and 300,000 frames recorded under surveillance camera.

@inproceedings{cheng2021rwf,
  title={Rwf-2000: An open large scale video database for violence detection},
  author={Cheng, Ming and Cai, Kunjing and Li, Ming},
  booktitle={2020 25th International Conference on Pattern Recognition (ICPR)},
  pages={4183--4190},
  year={2021},
  organization={IEEE}
}