Scholarlink

This study proposes a deepfake image detection model that combines MobileNetV2 with a frequency-domain analysis module to improve performance. The fusion of spatial and frequency features, along with an attention mechanism, enables the model to detect subtle manipulation artifacts. Experimental results demonstrate high accuracy and robustness, confirming the model’s effectiveness and efficiency for deployment on resource-limited platforms.

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

Detection of Deepfakes in Images Using Convolutional Neural Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Angela Paez-Barajas,
  • Daniela Cascavita-Mendieta,
  • Diego Renza

摘要

This study proposes a deepfake image detection model that combines MobileNetV2 with a frequency-domain analysis module to improve performance. The fusion of spatial and frequency features, along with an attention mechanism, enables the model to detect subtle manipulation artifacts. Experimental results demonstrate high accuracy and robustness, confirming the model’s effectiveness and efficiency for deployment on resource-limited platforms.