Impact Factor

7.883 Call For Paper

Share on

Author(s):

B. Nirmal | P. Abinesh | A. Manikandan | Mrs. K. Menaka
Keywords:

Interview Fraud Detection, Computer Vision, Online Proctoring, MediaPipe, TensorFlow.js, Firebase, Integrity Monitoring, Multi-Voice Detection, Web Application
Abstract:
The Rapid Shift To Remote Hiring Has Created New Avenues For Interview Fraud. This Paper Presents An Automated AI System For Interview Fraud Analysis And Alerts, A Browser-native Web Application That Monitors Online Interview Sessions In Real Time Using A Multi-modal Detection Pipeline. The System Integrates Computer Vision Via MediaPipe Face Mesh And TensorFlow.js COCO-SSD For Face Monitoring, Eye Gaze Estimation, And Object Detection; Web Audio API For Multi-voice And Whisper Detection; A QR Code-based Mobile Device Pairing Module Backed By Firebase Realtime Database; And Browser Tab-switch Detection Using The Page Visibility API. All Violations Are Mapped To A Dynamic Integrity Score Beginning At 100 Points. An Administrator Dashboard Provides Post-session Review With Candidate-level Integrity Scores And Violation Histories. The System Extends Prior Work On Audio-visual Synchronization-based Fraud Detection By Delivering Proactive, Real-time Detection Across Six Independent Modalities Entirely Within The Candidate's Browser.