Skip to content

A Deepfake Detection Project using EfficientNetV2 and FaceForensics++ with Gradio as UI

Notifications You must be signed in to change notification settings

rajobrata2003/deepfake-detection-efficientnet

Repository files navigation

Deepfake Detection

This repository provides code for detecting deepfake images and videos using pre-trained models.

Requirements

Run pip install -r requirements.txt to install all the required packages Download the model weights by running the download_weights.py script Make sure all codes and model weights are in same directory

Usage

Run the webui code depending on your requirement (Image or Video), upload the required media, and click on Submit Prediction will be generated accordingly.

Details

EfficientNet was used for the base feature extraction. A small portion of the FaceForensics++ dataset was used. Only around 8000 images were used for detection. Accuracy can be improved using more images and more variety.

About

A Deepfake Detection Project using EfficientNetV2 and FaceForensics++ with Gradio as UI

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages