Scholarlink

Artificial intelligence (AI) and deep learning (DL) have found applications across various domains, including automobiles, unmanned aerial vehicles (UAV), chatbots, military, and multimedia, offering numerous advantages. However, the rise of deepfake technology, a product of AI and DL, has introduced significant challenges. Originally employed in industries such as television, video games, and cinema to enhance visual effects, deepfakes are now being used for malicious purposes, including misinformation, harassment, blackmail, and sowing social discord. These AI-generated multimedia contents are highly realistic, making it difficult to differentiate between genuine and fake material. This research paper aims to analyze various techniques for detecting and classifying deepfake content, focusing primarily on machine learning and deep learning approaches. Since a variety of methods are used to create deepfake content, multiple deepfake datasets are employed for detection research. Popular tools like FaceSwap, DFaker, DeepFacelab, FaceSwap-GAN, and STGAN are commonly used to generate such content. With this purpose, the study explores 45 research articles that introduce various techniques for detection of deepfakes and evaluate their strengths and weaknesses. This insight becomes a severe consequence for countering this emerging threat from deepfakes.

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

A Survey on DeepFake Detection Using Machine Learning

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Santoshkumar Chobe,
  • Swati Nikam,
  • Aditya Bhosale,
  • Atharva Khairnar,
  • Kushal Patil,
  • Pritam Fulari

摘要

Artificial intelligence (AI) and deep learning (DL) have found applications across various domains, including automobiles, unmanned aerial vehicles (UAV), chatbots, military, and multimedia, offering numerous advantages. However, the rise of deepfake technology, a product of AI and DL, has introduced significant challenges. Originally employed in industries such as television, video games, and cinema to enhance visual effects, deepfakes are now being used for malicious purposes, including misinformation, harassment, blackmail, and sowing social discord. These AI-generated multimedia contents are highly realistic, making it difficult to differentiate between genuine and fake material. This research paper aims to analyze various techniques for detecting and classifying deepfake content, focusing primarily on machine learning and deep learning approaches. Since a variety of methods are used to create deepfake content, multiple deepfake datasets are employed for detection research. Popular tools like FaceSwap, DFaker, DeepFacelab, FaceSwap-GAN, and STGAN are commonly used to generate such content. With this purpose, the study explores 45 research articles that introduce various techniques for detection of deepfakes and evaluate their strengths and weaknesses. This insight becomes a severe consequence for countering this emerging threat from deepfakes.