Deep Convolutional Neural Network for Identifying Seam-Carving Forgery

Seam carving is a representative content-aware image retargeting approach to adjust the size of an image. To preserve visually prominent content, seam-carving algorithms first calculate connected path pixels, referred as seam, according defined cost function and then by removing or duplicating repeatedly calculated seams. actively exploited overcome diversity in resolution images between applications devices; hence, detecting distortion caused seam has become important forensics. In this paper, we propose convolutional neural network (CNN)-based classifying forgery. attain ability learn low-level features, designed (CNN) architecture comprising five types blocks specialized capturing local artifacts carving. An ensemble module further adopted both enhance performance comprehensively analyze features areas. validate effectiveness our work, extensive experiments based on various CNN-based baselines were conducted. Compared baselines, work exhibits state-of-the-art terms three-class classification (original, inserted, removed). The experimental results also demonstrate that model with robust for unseen cases.