Overview / Usage

Deep-Fake encompasses various technologies employed to create audio, image, and video hoaxes. The widespread accessibility of these tools has led to an unprecedented surge in AI-driven content generation. However , the unauthorized use of individuals' information to create such hoaxes raises significant concerns .Here we're tasked with devising techniques to detect Deep-Fakes, addressing the need to safeguard against the misuse of personal data in fabricated media content.

Methodology / Approach

Efficient Data Gathering : Our model employs Intel® oneAPI: DPC++ & oneTBB for swift acquisition from diverse sources like Social Media Feed Readers, Website Scrapers, and User Upload Modules.

Optimized Pre-processing: Leveraging Intel® oneAPI: MPG, IPP, DPC++ & oneTBB, our system ensures top-notch data quality through Format Conversion, Noise Reduction, and Segmentation.

Parallel Feature Extraction: With Intel® oneAPI: DPC++ & oneTBB, our model swiftly extracts Visual and Audio features in parallel, enhancing efficiency and speed.

Accelerated Deep Fake Detection: Utilizing Intel® oneAPI : oneDAL, our Deep Learning Model is trained efficiently, ensuring accurate detection of deepfakes with high confidence.

Accelerated Deep Fake Detection: Utilizing Intel® oneAPI: oneDAL, our Deep Learning Model is trained efficiently, ensuring accurate detection of deepfakes with high confidence. . e insights.

Reliable Protection: Backed by Intel®'s cutting-edge technology, our model offers reliable protection against deepfake manipulation, ensuring the integrity of your data and content.

Technologies Used

Programming Language : Python Frameworks : Tensorflow , Pytorch

Audio & video processing Libraries: Librosa , OpenCV

Machine Learning Tools: scikit-Learn , XGBoost, Vision Transformer Architecture

API : Intel OneAPI ,oneDAL

Web Technologies : HTML , CSS , JS, Django

Repository

https://github.com/Amar1701/deepfake_detection