Skip to content

[invasion_off_1]: TrustEngine - Multimodal Deepfake Detection #8

@AryanKhandare

Description

@AryanKhandare

Team Name

Team 404

Hackathon Track

AI/ML

ps-number

4

Email Address

aryankhandare2005@gmail.com

Email Addresses of Team Members

vrusha2107@gmail.com, bhaktirathod10@gmail.com, meetmak231@gmail.com

Project Description

We built a Multimodal Deepfake Detection & Trust Engine that analyzes videos to determine whether they are real or fake using AI.

The system allows users to upload media, processes it through a deep learning model, and returns a real-time authenticity score along with confidence levels.

It includes:

Video-based deepfake detection using frame-level analysis

Confidence scoring and risk classification

A web interface for easy upload and instant results

Backend API for integration with other platforms

Overall, it acts as a trust layer for digital media, helping users verify authenticity.

Inspiration behind the Project

With the rapid rise of generative AI, deepfakes are becoming more realistic and harder to detect. This creates serious risks like misinformation, identity fraud, and loss of trust in digital content.

We chose this problem because it has real-world impact — especially in social media, news, and financial systems.

Our goal was to build a system that not only detects deepfakes but also provides a confidence score, so users can make informed decisions instead of blindly trusting media.

We focused on making the solution practical, scalable, and easy to use, rather than just a research model.

Tech Stack

FRONTEND
React → building user interface
Vite → fast development & project setup
Tailwind CSS → styling and responsive design
Axios → sending API requests to backend

BACKEND
FastAPI → creating API for prediction
Uvicorn → running backend server

MACHINE LEARNING
PyTorch → building and running deep learning model
ResNet18 → model for deepfake detection
Torchvision → image transformations & model utilities

DATA PROCESSING
OpenCV → video reading & frame extraction
NumPy → numerical operations (averaging scores)

DATASET
FF++ / Celeb-DF → training data for real vs fake videos

Project Repo

https://github.com/AryanKhandare/Trust_Engine.git

Demo Video

https://drive.google.com/file/d/1Raq2bOGUseUr4vRbliiTscIWcEOo5-ao/view?usp=sharing

Presentation Link

https://canva.link/3ng7obts8vb9ds3

Anything Else?

NA

Rules and Code of Conduct

  • I agree to follow this hackathon's Rules and Code of Conduct

Metadata

Metadata

Assignees

No one assigned

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions