Scholarlink

It is an important aspect of digital media proofing that ensures the credibility and availability of content all over the world. This study implemented the RESNET50 framework, by which forgery videos in a set containing 20 sequences (10 originals and the other 10 themselves forgeries) were identified. Finally, there was the standardization process comparing four files that were shot at 30 frames per second with 320 × 240 (SULFA database) pixel resolution, and coded through various conversion and compression processes. This abandoned sequence file and MAT file offers all the variations in Y, U, and V components description and its significance. The performance of our model as given by our technique improves to 80% accuracy proving; thus, the ability of RESNET50 in identifying any change that happens is videos. This study will boost the development of video forgery detection techniques effective in challenging circumstances, comprising various compression formats and resolutions.

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

Advancements in Video Forgery Detection Using Temporal Residual Networks: A Deep Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Hemant A. Tirmare,
  • Jaydeep B. Patil

摘要

It is an important aspect of digital media proofing that ensures the credibility and availability of content all over the world. This study implemented the RESNET50 framework, by which forgery videos in a set containing 20 sequences (10 originals and the other 10 themselves forgeries) were identified. Finally, there was the standardization process comparing four files that were shot at 30 frames per second with 320 × 240 (SULFA database) pixel resolution, and coded through various conversion and compression processes. This abandoned sequence file and MAT file offers all the variations in Y, U, and V components description and its significance. The performance of our model as given by our technique improves to 80% accuracy proving; thus, the ability of RESNET50 in identifying any change that happens is videos. This study will boost the development of video forgery detection techniques effective in challenging circumstances, comprising various compression formats and resolutions.