En-VStegNET: Video Steganography using spatio-temporal feature enhancement with 3D-CNN and Hourglass

摘要: Learning Spatio-temporal features has shown improved performance on tasks involving video analysis using deep learning, and the learning community used these to solve a varied variety of problems. Video steganography is one such problem where for can help improve steganography. Steganography practice concealing confidential information, protect information from an adversary, into ordinary cover message in way that does not seem suspicious adversary. Recent deep-learning-based methods have proven secrecy capacity over traditional techniques. In this paper, we propose novel state-of-the-art 3D-CNN architecture with enhancement feature full The proposed model outperforms current both qualitatively quantitatively. We validated our by comparing it new as well techniques, quality different statistical metrics, namely, PSNR, SSIM, APD, VIF at frame, level. Moreover, check undetectability model, subjected detection steganalysis tools like SRNet. Results fine-tuning classifiers, ResNet Inception-v3, detect steganographic messages maintains model’s accuracy.