Towards Prototyping Smart Door System Based on Facial Recognition

Project Title

Towards Prototyping Smart Door System Based on Facial Recognition

Project Area of Specialization Artificial IntelligenceProject Summary

Security issues has been of primary concern in recent years. Amongst many environments, home security is the most vulnerable as it is managed by individuals with limited resources. Older security systems can’t adapt the new technology as well as various bugging methods for sabotaging the individual’s or family protection. With recent news on intruders and theft, it is necessary that an efficient and effective system is designed which can adapt new technologies as well as cope with the increasing vulnerabilities of security system. In this regard, a smart home security system based on IoT analytics is proposed which will consider face as its primary biometric trait and also requires administrative assistance for authorizing new users. The face recognition system is based on machine learning to make the system reliable and efficient. The use of such techniques does not require heavy machinery or sophisticated infrastructure which also suits the needs of an individual, hence, making the system more effective. The decision from the face recognition system will be sent to a the designated security system which can not only perform the authorization but also can send messages/information to the home owner for intruder alert.

Project Objectives

• Aims:The main aim of this project is to design an efficient machine learning algorithm based system for face recognition and evaluate its performance.
• Objectives of Project: The main objectives of this project are as follows:
  • Design a face recognition algorithm using machine learning techniques.
  • Perform comparative analysis of the designed face recognition algorithm with Microsoft Azure.
  • Integration with Raspberry Pi 3 Model B+ to build a prototype and make it compliant with IoT systems.

Project Implementation Method

Methodology

1. Image Acquisition:

Image acquisition accomplished by digitally scanning an existing photograph or by live Webcam.

2. Pre-Processing / Normalization:

Face recognition algorithms have to deal with significant amounts of illumination for enhancing image quality.

3. Feature Extraction / Encoding:

Image after pre-processing fed to a feature extraction scheme to extract features from it.

4. Face Recognition: Its divided in two separate stages :

1. Training Process : where the algorithm is fed samples of the subjects to be learned and a model is determined.
2. Evaluation process : where a image of a test subject is tested against trained  model and the decision is drawn whether the subject is authorized or unauthorized .

Benefits of the Project

Facial recognition – Many smart home systems allow to view your kids entering the front door when they get home. 

Smart locks – Everyone forgets to lock the door behind them from time to time.

Mobile alerts – Smart systems allow you to set up alerts for almost anything. Get notified when someone opens a door or window

Remote accessibility – Forgot to arm the house? No problem. Like your locks, smart systems allow you to “arm” from anywhere.

Technical Details of Final Deliverable

• Hardware

1. Raspberry Pi 3 Model B+
2. HD Webcam
3. GSM module
4. Stepper motor / Servo Motor
5. Monitor Display
6. Relay and Power Supply
7. Ledz , PIR and Buzzer sensor
8. Breadboard, Push Buttons and Connecting Wires

• Software

1. Microsoft Visual Studio 2015
2. Microsoft Face Api
3. Windows 10 Iot Dashboard Manager
4. MATLAB R2018/Python

Final Deliverable of the Project HW/SW integrated systemType of Industry IT , Security Technologies Artificial Intelligence(AI), Internet of Things (IoT)Sustainable Development Goals No Poverty, Zero Hunger, Reduced InequalityRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	35000
Total items Expenditure	Equipment	1	35000	35000