The problem Securing remote hiring through Deepfake Detection solves

The increasing adoption of online interviews in digital recruitment has introduced significant security challenges. Advanced deepfake technologies enable the real-time manipulation of video and audio, allowing individuals to impersonate candidates with high realism.

Current interview platforms lack robust mechanisms for real-time detection and continuous verification of candidate authenticity. As a result, these systems are vulnerable to identity fraud, impersonation, and unauthorized participation.

This limitation compromises the integrity of the hiring process, leading to unreliable candidate evaluation, potential security risks, and reduced trust in remote recruitment systems.

Challenges we ran into

Specific hurdle: Deepfake detection models typically require large datasets and long training times, making it difficult to achieve high accuracy within limited hackathon time.

Solution: We propose a real-time deepfake detection system that analyzes video input to identify manipulated content. The system captures live video, extracts frames, and detects faces. These faces are processed using a trained deep learning model (EfficentNet,Inceptionresnet) to identify inconsistencies like unnatural facial movements, blinking patterns, and texture irregularities and also gives confidence score.

The model classifies the video as real or fake in real time. For higher accuracy, audio analysis and lip-sync verification can also be integrated. The system can be deployed in online interview platforms to prevent identity fraud and ensure authenticity.