Deep Learning Hierarchical Representations for Image Steganalysis

By Jian Ye, Jiangqun Ni, Yang Yi. IEEE Transactions on Information Forensics and Security, 2017

736 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Yedroudj-Net: An Efficient CNN for Spatial Steganalysis (2018, methodology): “…of the CNN, avoiding the use of tricks known to improve the performances such as transfer learning [16] or virtual augmentation of the database [17], etc. Additionally, the proposed network is not sensitive to the initialization of hyperparameters and thus easily converges, which will be later…”
Depth-Wise Separable Convolutions and Multi-Level Pooling for an Efficient Spatial CNN-Based Steganalysis (2020, methodology): “so-called selection channel [43], [44]) in [19] and [39], the awareness of JPEG phase in [18] and [45] (the underlying grid of 8 × 8 pixels [46]) for JPEG steganalysis.”
Deep Residual Network for Steganalysis of Digital Images (2019, methodology): “The next advancement, the YeNet [65], can be considered as a breakthrough result as the proposed detector significantly improved upon established steganalysis detectors in the spatial domain.”
SteganoGAN: High Capacity Image Steganography with GANs (2019, methodology): “Finally, we compare the detection error for images generated by STEGANOGAN against those reported by (Ye et al., 2017) on images generated by three state-of-the-art steganography algorithms: WOW (Holub & Fridrich, 2012), S-UNIWARD (Holub et al., 2014), and HILL (Li et al., 2014).”