Face-PAD-GUI
A GUI based face spoof detector based on spoof detection algorithm with explainability traces.
Project Repository
In safety-critical biometric systems, spoof detection is crucial. Individuals may attempt to evade detection or acquire documents, such as passports or other legal IDs, by posing as someone else. Face spoof detection helps prevent attempts to obscure or falsify identity, ensuring the integrity and reliability of biometric verification systems.
Key Applications of Iris Spoof Detection:
- Border Security and Immigration: Ensures that travelers are accurately identified, preventing unauthorized entry and impersonation attempts at border control points.
- Identity Verification for Legal Documents: Helps in verifying the authenticity of identities for passports, driver’s licenses, and other legal documents, ensuring that only legitimate holders can obtain these documents.
- Secure Access Control: Used in high-security facilities, such as government buildings, research labs, and financial institutions, to prevent unauthorized access through identity spoofing.
- Financial Transactions: Protects against fraudulent activities in financial institutions by verifying genuine users through iris recognition, reducing identity theft in banking and online transactions.
- Healthcare Access: Confirms patient identity in secure health systems, ensuring that only authorized individuals receive sensitive medical information and services.
Key Features
- Built in face-detector based on mtcnn for face detection.
- Face spoof detection. Score output is between [0, 1].
- Traces of spoof on all four kinds of traces.
- Single image processing mode and batch processing mode.
---
layout: page
title: project
description: a project with a background image
img: /assets/img/12.jpg
---
Application Screenshots:
Some of the screenshot of the application. Score: 0: Bonafide, and 1: Presentation Attack (PA). The first one is a bonafide face which has a very low PA score (0.08). The second one is a transparent mask which has high PA score (0.83) and the third one has a printed face mask which also exhibits high PA score (0.45).
Case 1:
The input here is a real image and the spoof traces on the 2x2 grid on the right side shows a very little activity.
Case 2:
The input here is a real face with transparent masks with extra eye-brows and mostache. This images shows high PA score and also the traces on the right side shows what are the regions primary contributing to the decision.
Case 3:
This face image is a real face with printed face mask on. This also shows a high PA score and the right side shows the traces of the spoof.