Argus
A DeepFake Detection WebApplication which detects DeepFakes
Created on 16th March 2024
•
Argus
A DeepFake Detection WebApplication which detects DeepFakes
The problem Argus solves
Recently we have ran into many face forgery crimes, faces of innocent people have been forged and misused as deepfakes . As the technology has been increasingly it is difficult to identify whether the image or video is fake or real. So, we came up with a solution of implementing a deep learning model to identify the deepfakes using some high-level models like MesoNet . It analyzes the image by extracting its facial features and then classifies it as fake or real.
Challenges we ran into
As this is a huge project we have difficulties in the datasets because the datasets size were around 450GB to 3.5TeraBytes . So we chose another smaller dataset which consists of 20,000 images and trained it . Due the dataset size we were unable to get em on Google Cloud or Azure , So we had to cut it down on our local GPU's.
Tracks Applied (1)
Developer Productivity
Technologies used