Scholarlink

Deep learning techniques have found successful applications across various domains, including image detection, big data analytics, voice and image recognition. Deepfakes, a product of combining deep learning with fake creation, involve generating deceptive images or videos using AI, presenting risks such as political manipulation, misinformation, and exploitation. This has raised pressing concerns related to privacy, security, and ethical implications. Notably, face-swapping deepfake methods are prevalent, producing highly realistic videos that pose serious threats to individual and national privacy. Consequently, discerning between authentic and deepfake videos has emerged as a severe challenge.This research focuses on solving this challenge using MTCNN model (Multitask Cascaded Convolution Networks) giving accuracy rate of 93.7%.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deep Fake Detection: Integration of Inception-Net and Resnet Multitask Cascaded Convolution Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Rajvardhan Nalawade,
  • Bhanudas Hegaje,
  • Bhargav Maiskar,
  • Prerana Patil,
  • Uma Gurav

摘要

Deep learning techniques have found successful applications across various domains, including image detection, big data analytics, voice and image recognition. Deepfakes, a product of combining deep learning with fake creation, involve generating deceptive images or videos using AI, presenting risks such as political manipulation, misinformation, and exploitation. This has raised pressing concerns related to privacy, security, and ethical implications. Notably, face-swapping deepfake methods are prevalent, producing highly realistic videos that pose serious threats to individual and national privacy. Consequently, discerning between authentic and deepfake videos has emerged as a severe challenge.This research focuses on solving this challenge using MTCNN model (Multitask Cascaded Convolution Networks) giving accuracy rate of 93.7%.