Patch, Spectral and Eye Pattern based Deepfake Video Detection

Deep fake detection is necessary for several reasons: Misinformation: Deepfakes have the potential to disseminate misleading information and influence public perception, posing significant risks to individuals, organizations, and society as a whole. Deception: Deepfakes can be used to deceive people and manipulate their perceptions, leading to distrust in the authenticity of media. Privacy invasion: Deepfakes can be used to defame individuals and invade their privacy, causing emotional distress and harm to their reputation. Security: Deepfakes can be used in political and security contexts to spread false information and incite violence. Therefore, the need for deep fake detection stems from the need to maintain trust in the authenticity of media and to prevent the malicious use of deepfakes for deceptive or harmful purposes. Deepfake detection spans multiple areas, particularly in image and video analysis, where techniques are developed to identify anomalies and inconsistencies. This includes monitoring irregularities in facial expressions, eye movements, and body language. The proposed system aims in detecting fake videos generated using deep learning techniques by analysing different anamoly pattern such as change in pixelvalues, energy band analysis, spectrogram analysis and eye pattern using different machine learning techniques.