First Steps Toward Camera Model Identification with Convolutional Neural Networks

TL;DR

This paper introduces a convolutional neural network-based method for camera model identification that learns directly from images, outperforming existing algorithms and generalizing to unseen camera models.

Contribution

It presents a novel data-driven CNN approach for camera model identification that surpasses state-of-the-art methods and generalizes to new camera models.

Findings

Outperforms current algorithms on 64x64 image patches

Features generalize to unseen camera models

CNN learns distinctive camera-specific features

Abstract

Detecting the camera model used to shoot a picture enables to solve a wide series of forensic problems, from copyright infringement to ownership attribution. For this reason, the forensic community has developed a set of camera model identification algorithms that exploit characteristic traces left on acquired images by the processing pipelines specific of each camera model. In this paper, we investigate a novel approach to solve camera model identification problem. Specifically, we propose a data-driven algorithm based on convolutional neural networks, which learns features characterizing each camera model directly from the acquired pictures. Results on a well-known dataset of 18 camera models show that: (i) the proposed method outperforms up-to-date state-of-the-art algorithms on classification of 64x64 color image patches; (ii) features learned by the proposed network generalize to camera models never used for training.