ILLUMINATING DEEPFAKES - An Anti-Deepfake TechnologyBy Team RnD 4 Dr. D.Y.PATIL SCHOOL OF ENGINEERING PUNE DEPARTMENT OF COMPUTER ENGINEERING Presented by: R&D Group 4 Team Members 1. Mrityunjay Mishra PRN No.71920213c 2.Shaikh Azim PRN No.71920289C 3.Allaudin Shaikh PRN No.71920287G 4.AnandKumar Rai PRN No.71845962J UNDER THE GUIDANCE OF Prof. Sunil Rathod Area of Project: A. I & Digital Forensics CONTENTS Track 1 PROBLEM STATEMEMT MOTIVATION OBJECTIVES INTRODUCTION LITERATURE REVIEW PROJECT FEASIBILITY & Scope REQUIREMENTS ARCHITECTURE CONTENTS Track 2 ALGORITHMS STRUCTURAL DIAGRAMS BEHAVIORAL DIAGRAMS ADVANTAGES LIMITATIONS FUTURE WORK CONCLUSION REFRENCE PROBLEM STATEMENT 120 100 100 90 80 80 70 60 50 40 20 0 Percentage Due to an advent rise in technology, it has almost become very difficult to distinguish between deepfakes and real video. Deepfake videos are usually made by using StyleGAN’s(Generative Adversial Network) framework and other deep learning technologies. The danger of this is that technology can be used to make people believe something is real when it is not actually. For finding a secured solution for this we need to develop an antideepfake technology that will be implementing deepfake detection models with different features using a deep learning approach to detect these videos by isolating, analyzing, and verifying the content. MOTIVATION 120 100 100 90 80 80 70 60 50 40 20 0 Percentage Few applications such as FaceApp and Fake App are built on this technology to be used on mobile and desktop, which results in affecting a person’s integrity. Identifying, authenticating, and categorizing these videos has become a necessity. It’s not just about these latest advancements in creating fake images and video, it is the injection of these techniques into an ecosystem that is already promoting fake news, sensational, and conspiracy theories. OBJECTIVE 90% CNN WILL BE USED TO CLASSIFY REAL OR FAKE AMONGST ALL. DEDIDACETD FUNCTIONS IMAGE EXTRACTIONVIDEOS WILL BE CONVERTED INTO FRAMES EXPLOITING FRAME-LEVEL SCENE INCONSISTENCY 10% 20% 45% 75% EXPLOITING MULTIPLE ANOMALIES & LEADS TO A FLICKERING PHENOMENON 50% COMPARING EXCLUSIVE BLUR INCONSISTENCY BY THE SYNTHESIZED BLURRED AREAS INTRODUCTION TO DOMAIN ARTIFICIAL INTELLIGENCE . AI which stands for artificial intelligence refers to systems or machines that mimic human intelligence to perform tasks and can iteratively improve themselves based on the information they collect.AI isn’t intended to replace humans. It’s intended to significantly enhance human capabilities and contributions. That makes it a very valuable business asset. DIGITAL FORENSICS Digital forensics is a branch of forensic science encompassing the recovery, investigation, examination and analysis of material found in digital devices, often in relation to mobile devices and computer crime. Machine learning is a branch of artificial intelligence (AI) and computer science which focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy LITERATURE REVIEW PAPERS Deepfake Video Detection Using RNN AUTHOR David Guera METHODS Video and image processing Laboratory Exposing Deepfake videos By Detecting Warping Artifacts Yuezun Li Eyes blinking, Head poses using Warping Artifacts The Deepfake Detection Challenge Dataset Brian Dolhansky Audio Visual affective cues Deepfake Video Detection Through Optical Irene Amerint Flow Based CNN CNN, Generated imagess UPDATION ADDITION SCOPE Adding additional images to the dataset, as having more data will inherently make the models more accurate. Update the associated WebApp to be more user-friendly and have more features for deepfake image classification. EXPANSION FACEFORENSICS Expand the scope of this project to include being able to scan video for deepfakes as this is increasingly becoming problematic technology. It consists of an encoder and a decoder, where at the encoder side, the image is given as input, and scaling is done using a convolutional layer REQUIREMENTS Processor : 2.0Gz HARD DISK: 128 GB RAM : 8 GB OS: WINDOWS 7 OR ABOVE LANGUAG E: PYTHON 3, html High end integrated tools and extensions SYSTEM ARCHITECTURE STRUCTURAL ARCHITECTURE CLASS DIAGRAM BEHAVIORAL ARCHITECTURE DOWNLOAD THE DATASET Step #1 OPTIMIZATION AND TRAINING WITH CNN FACE DETECTION Step #2 ENCODING LABELS Step #3 Step #4 READING GREYSCALES & SPLITTING DATA PROCESSES Step #5 ADVANTAGES & LIMITATIONS ADVANTAGES: • Machine Learning is one of the accurate techniques that can be applied in the system for detecting deepfakes. • We all know deepfakes are difficult to identify with our naked eyes, so human sight error can be eliminated. • User-Friendly interface. • Effective Prediction System. LIMITATIONS: • One of the most important challenges facing the researchers is the lack of high-quality dataset. • deep learning models often require large dataset for the training step in order to produce good results, which are not freely accessible or need permission from social media providers. • The rapid development of deepfake GAN models can also bring a new challenge where unseen types of generated images and video may not be discovered by the current deep learning models APPLICATIONS Using Machine Learning models, we will be able to predict whether the given data is fake or real. The model firsts use a deep learning network to extract face features based on face recognition networks. Then, a fine-tuning step is used to make face features suitable for real/fake image detection. This deepfake image detection system can be used in many sectors, including social media companies, security organizations, and news agencies FUTURE WORK 1. Improving Learning videos/images. model for better prediction of fake 2. Although deep learning has shown a remarkable performance in deepfakes detection, the quality of deepfake has been increasing. Hence, the current deep learning methods need to improve as well to successfully identify fake videos and images 3. The current deep learning methods, there is not a clear method to know the number of layers needed and which architecture is appropriate for deepfake detection. CONCLUSION Deepfake had become popular due to the massive availability of images and videos in social content. This is particularly important nowadays because the tools for making deepfakes are becoming more accessible, and social media sites will easily allow people to distribute and share such fake content. In this project, we describe a new deep learning-based method that can effectively distinguish AIgenerated fake videos (DeepFake Videos) from real videos. Our method is based on the observations that the current DeepFake algorithm can only generate images of limited resolutions, which are then needed to be further transformed to match the faces to be replaced in the source video. Such transforms leave certain distinctive artifacts in the resulting DeepFake Videos, which can be effectively captured by a dedicated deep neural network model. We evaluate our method on several different sets of available DeepFake Videos which demonstrate its effectiveness in practice. REFRENCES [1]https://ayushbasral.medium.com/deepfake-detection-using-resnxt-and-lstm-bcc08c086f84. [2]https://lab.irt.de/a-system-for-deepfake-detection-dfirt/ . [3]FaceForensics++: Learning to Detect Manipulated Facial Images by Andreas Rossler ¨ 1 Davide Cozzolino2 Luisa Verdoliva2 Christian Riess3 Justus Thies1 Matthias Nießner1. [4]Deepfake Video Detection Using Recurrent Neural Networks David Guera Edward J. Delp ¨ Video and Image Processing Laboratory (VIPER), Purdue Universit. [5]Digital Forensics and Analysis of Deepfake Videos Mousa Tayseer Jafar Mohammad Ababneh Mohammad Al-Zoube Ammar Elhassan. [6] Review of Deepfake Detection Techniques by M P Adithya Vijayan, Thushara P, Sanjana S, Karthik P C. [7] Deepfake Detection using ResNxt and LSTM _ by Ayush Basral _ Medium. Thank You