Gestural Interaction and Facial based Authentication System
The first step is to swipe the debit card for the transaction. The system will check the validity of the card such as its expiry da
2025-06-28 16:32:44 - Adil Khan
Project Title
Gestural Interaction and Facial based Authentication System
Project Area of Specialization Artificial IntelligenceProject Summary- Authentication systems play a vital role in securing our credentials for social, personal, and monetary platforms. The Uni-modal systems, i.e. based on a single modality, has proven to be ineffective due to its compromising nature. Most of the monetary transactions are only based on debit, credit cards using PIN or OTP (One Time Password). With the advancement of vision system, multifactor authentication can be added on top of the PIN & OTP authentication layer to make the transaction more secure.
- In this project, we consider The Gestural Interaction and Facial based Authentication System along with PIN or OTP to authorize the transaction.
- To survey the existing works using face and gesture modalities.
- To collect facial images and gesture-based passwords from multiple users.
- To evaluate video representation methods for summarizing gestural videos in a single image such as motion history images, dynamic images, and optical flows.
- To employ machine learning specifically deep learning approached for designing multi-modal authentication system.
- The first step is to swipe the debit card for the transaction.
- The system will check the validity of the card such as its expiry date. If the card is valid the system will proceed to next step or else will withdraw the transaction immediately.
- The system will then validate the PIN or OTP provided by the user. The system will proceed if the PIN or OTP is validated and verified.
- The GIAFAS adds the multifactor authentication once the PIN or OTP is verified. Now GIAFAS will checks that user of debit card is the real user or not through Gestural Interaction and Facial based Authentication. Each of these components are briefly explained below:
- The Gestural interaction: The gestural interaction is categorized as behavioral trait in this project. Multiple videos from the user for the same password will be recorded in the form of videos. We will employ one of the video summarizing representation as suggested in the literature. We then train machine/deep learning model for training the gestural passwords to recognize the user.
- Facial recognition: furthermore we will add the facial authentication method to provide more security for the transaction, the system will recognize the face of debit user using FaceNet model or we may train our own network to recognize the face as well.
- Once the gestural interaction and facial recognition results are available a decision-level fusion will be employed to combine the results and draw the final conclusion, i.e. authorized or not-authorized, accordingly. The transaction will be processed if the result is authorized and vice-versa.
- A desktop application as well as website will be designed for attaining the said task.
- There is no need of physical interaction with the user after pin code verification
- Only single device will be required for facial and gesture based recognition.
- The system incorporates one physical and one behavioral trait so that if the physiological one gets compromised the behavioral trait can easily be changed.
- Desktop Application and Web Application for Gestural Interaction and Facial based Authentication System for the developing these systems we will use Python (PyQt or other framework for win form) and Django (web form) programming languages and also we use different algorithms and packages for regonizing face and gesture authentication like FaceNet, OpenCV, Tensorflow, and PyTorch, Scikit learn.
- Motion Represented images for gestural videos using MHI, dynamic images or the optical flows.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 50563 | |||
| Intel Real Sense Depth Camera D435 | Equipment | 1 | 27511 | 27511 |
| AstraPro2018R3 | Equipment | 1 | 23052 | 23052 |