Problem

Social media platforms have become a breeding ground for fake news and deepfake videos, which can cause significant harm to individuals and society. These videos can be used to spread misinformation, defame someone's character, or even manipulate public opinion. Traditional methods of detecting deepfakes are not always effective, making it challenging to prevent their spread.

Our team was tasked with developing a cutting-edge deepfake detection system that could accurately identify these videos and prevent them from spreading on social media platforms.

Learn more about our solution here.

Solution

We developed a deepfake detection system using a combination of Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) approaches. Our system analyzed the Celeb-DF (v2) dataset by extracting and cropping faces from videos to train the model.

The LSTM model was used to analyze the temporal features of the video, while the CNN model focused on analyzing spatial features. By combining these two models, we were able to achieve an accuracy rate of 95.33% in deepfake video classification.

Impact

Reduced the spread of fake news and misinformation on social media platforms
Increased public trust in online content by providing an effective way to detect deepfakes
KPI Placeholder: Improved accuracy rate in deepfake video classification by X%

Quote of how incredible of an impact this had.

Team

Akash Shekhavat
- Developed the LSTM and CNN models for deepfake detection
- Analyzed the Celeb-DF (v2) dataset to train the mode

Developing Deepfake Detection System with 95.33% Accuracy