Scholarlink

This survey paper reviews the current advancements in deepfake detection technologies, focusing on improved datasets and algorithms. Deepfake technology, powered by AI and machine learning, poses significant risks, including misinformation and identity theft, necessitating robust detection mechanisms. The paper outlines the historical evolution from traditional forensic methods to sophisticated deep learning techniques like CNNs, RNNs, and transformers. Key milestones include the development of comprehensive datasets such as FaceForensics++ , DFDC, and Celeb-DF, which have significantly enhanced detection accuracy. Advanced techniques like transfer learning, data augmentation, and adversarial training are also discussed. Practical applications in social media and digital forensics are highlighted, demonstrating the effectiveness of these innovations. The survey concludes by emphasizing the need for continuous research and development to address emerging challenges in deepfake generation and ensure the integrity of digital content.

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 Accessible Deepfake Detection on Improved Datasets and Algorithms

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Dhammjyoti Dhawase,
  • Narendra Rathore,
  • Prajyot Panmand,
  • Vaishnavi Yadav,
  • Paras Patil

摘要

This survey paper reviews the current advancements in deepfake detection technologies, focusing on improved datasets and algorithms. Deepfake technology, powered by AI and machine learning, poses significant risks, including misinformation and identity theft, necessitating robust detection mechanisms. The paper outlines the historical evolution from traditional forensic methods to sophisticated deep learning techniques like CNNs, RNNs, and transformers. Key milestones include the development of comprehensive datasets such as FaceForensics++ , DFDC, and Celeb-DF, which have significantly enhanced detection accuracy. Advanced techniques like transfer learning, data augmentation, and adversarial training are also discussed. Practical applications in social media and digital forensics are highlighted, demonstrating the effectiveness of these innovations. The survey concludes by emphasizing the need for continuous research and development to address emerging challenges in deepfake generation and ensure the integrity of digital content.