AI based Django application to locate Missing People using Facial Features

A Django based age invariant face recognition project. In an effort to file report and find missing people with the joint help of the community.

Functionalities

• File Missing Cases
  • Images are verified for faces and preprocessed using MTCNN (Multi-task Cascaded Convolutional Networks developed for both face detection and face alignment)
• Match found person with missing people in database
  • MTCNN: is again used to detect face in the uploaded image
  • VGG16: is a type of CNN (Convolutional Neural Network) used for object detection and classification algorithm. Classify 1000 images of 1000 different categories with 92.7% accuracy. Used here for extracting features from last 3 layers and transfer learning.
  • PCA (Principal Component Analysis): is a dimensionality-reduction method that is often used to reduce the dimensionality of large data sets, by transforming a large set of variables into a smaller one that still contains most of the information in the large set.
  • MDCA (multimodal discriminant correlation analysis): a cascade version of DCA (Discriminant correlation analysis) maximize the correlation of corresponding features across the two input feature sets (as done in CCA-methods) in addition to performing a correlation for features that belong to different classes within each feature set. In addition to feature-level fusion, DCA performs significant dimension reduction, which implies low computational complexity.
• Admin Privilege
  • Delete records from database
  • View user activity
• Live dashboard for:
  • Total people missing, people found, users registered
  • Details of people found
  • Users Leaderboard
• Mail: account activation, reset password, remind username, found missing person
  • Courier: A multi-channel notifications service that allows you to bring your own email provider, including support for SMTP and most popular transactional email APIs.
• Multilingual: English, Hindi, Kannada.

The following basic features were forked from egorsmkv GitHub

• Log in & Log out
• Create an account
• Reset password
• Send/ Resend activation code
• Remind a username
• Change password
• Change profile

Requirements

• Anaconda
• Python
• Django
• Tensorflow
• Octave (Free Open source alternative to read matlab files)

Demo

Note: The images of people used belong to the FGNet Dataset.

Basic Features

django_locate_demo_basic.mp4

Account Activation

Forgot Password / Username

Find match with Missing People in the database

django_locate_demo_found_person2.mp4

When the missing person is found, an email is sent to Admin and the user who filed the FIR

File a new missing case

django_locate_demo_file_missing.mp4

Project Structure

locate-missing-people-ai-django
├─ .git
├─ .gitignore
├─ conda-requirements.txt
├─ dcaFuse.m
├─ fuse.m
├─ LICENSE
├─ README.md
├─ requirments.txt
├─ screenshots
│  ├─ authorized_page.png
│  ├─ create_an_account.png
│  ├─ login.png
│  ├─ password_change.png
│  ├─ password_reset.png
│  └─ set_new_password.png
└─ source
   ├─ accounts
   │  ├─ admin.py
   │  ├─ apps.py
   │  ├─ detect_face.py
   │  ├─ face_recog.ipynb
   │  ├─ FeatureExtraction.ipynb
   │  ├─ forms.py
   │  ├─ match_face.py
   │  ├─ migrations
   │  ├─ models.py
   │  ├─ templates
   │  │  └─ accounts
   │  │     ├─ emails
   │  │     │  ├─ activate_profile.html
   │  │     │  ├─ activate_profile.txt
   │  │     │  ├─ change_email.html
   │  │     │  ├─ change_email.txt
   │  │     │  ├─ forgotten_username.html
   │  │     │  ├─ forgotten_username.txt
   │  │     │  ├─ restore_password_email.html
   │  │     │  └─ restore_password_email.txt
   │  │     ├─ log_in.html
   │  │     ├─ log_out.html
   │  │     ├─ profile
   │  │     │  ├─ change_email.html
   │  │     │  ├─ change_password.html
   │  │     │  ├─ change_profile.html
   │  │     │  ├─ file_missing.html
   │  │     │  ├─ found_missing.html
   │  │     │  ├─ match.html
   │  │     │  ├─ missing_list.html
   │  │     │  └─ user_list.html
   │  │     ├─ remind_username.html
   │  │     ├─ resend_activation_code.html
   │  │     ├─ restore_password.html
   │  │     ├─ restore_password_confirm.html
   │  │     ├─ restore_password_done.html
   │  │     └─ sign_up.html
   │  ├─ urls.py
   │  ├─ utils.py
   │  ├─ views.py
   │  └─ __init__.py
   ├─ app
   │  ├─ conf
   │  │  ├─ development
   │  │  │  ├─ settings.py
   │  │  │  └─ __init__.py
   │  │  ├─ production
   │  │  │  ├─ settings.py
   │  │  │  └─ __init__.py
   │  │  └─ __init__.py
   │  ├─ settings.py
   │  ├─ urls.py
   │  ├─ wsgi.py
   │  └─ __init__.py
   ├─ content
   │  ├─ assets
   │  │  ├─ css
   │  │  ├─ favicon.png
   │  │  ├─ js
   │  │  └─ vendor
   │  │     ├─ bootstrap
   │  │     ├─ jquery
   │  │     └─ popper
   │  ├─ locale
   │  │  ├─ hi
   │  │  └─ kn
   │  └─ templates
   │     ├─ layouts
   │     │  └─ default
   │     │     └─ page.html
   │     └─ main
   │        ├─ change_language.html
   │        └─ index.html
   ├─ main
   │  ├─ apps.py
   │  ├─ views.py
   │  └─ __init__.py
   └─ manage.py

Install dependencies & activate virtualenv

For this project Anaconda environment is used. Dependencies are mentioned in conda-requirements.txt and requirements.txt

conda create -n tf_gpu_env -c conda-forge cudatoolkit cudnn python=3.8
        or
conda create -n tf_gpu_env python=3.8        
conda activate tf_gpu_env

Configure the settings (connection to the database, connection to an SMTP server, and other options)

1. Edit source/app/conf/development/settings.py if you want to develop the project.

2. Edit source/app/conf/production/settings.py if you want to run the project in production.

Apply migrations

python source/manage.py migrate

Collect static files (only on a production server)

python source/manage.py collectstatic

Running

A development server

Just run this command:

python source/manage.py runserver

Reference

Age-invariant face recognition based on deep features analysis
Paper: Moustafa, A.A., Elnakib, A. & Areed, N.F.F. Age-invariant face recognition based on deep features analysis. SIViP 14, 1027–1034 (2020). https://doi.org/10.1007/s11760-020-01635-1

FGNET Dataset Description
Homepage: https://yanweifu.github.io/FG_NET_data/
Paper: Yanwei Fu, Timothy M. Hospedales, Tao Xiang, Jiechao Xiong, Shaogang Gong, Yizhou Wang, and Yuan Yao. Robust Subjective Visual Property Prediction from Crowdsourced Pairwise Labels. IEEE TPAMI 2016
Point of Contact: y.fu@qmul.ac.uk