Detection-Aware ROI-Based Video Steganography Using Dual-Network Training
Loading...
Authors
Susoglu, Arzu
Issue Date
2026-04-17
Type
Other
Language
en_US
Keywords
Dual network training , Video steganography
Alternative Title
Abstract
This project proposes a detection-aware video steganography framework that integrates region-of-interest (ROI)-guided embedding with dual-network training. Objective: Improve imperceptibility and resistance to detection simultaneously
Approach: Incorporate a steganalysis network during training for adaptive embedding
Goal: Minimize detectability while preserving payload integrity and visual quality
Hypothesis: Detection-aware training can reduce steganographic detectability while preserving visual quality.
Significance: This work addresses vulnerabilities of modern steganography against deep learning-based detection.
Description
UB Rise 2026
Department of Computer Science and Engineering