MMFusion: Combining Image Forensic Filters for Visual Manipulation Detection and Localization

TL;DR

This paper introduces MMFusion, a novel architecture that combines multiple forensic filters to improve the accuracy of image manipulation detection and localization, outperforming existing methods.

Contribution

The paper proposes a new fusion-based architecture, MMFusion, that leverages complementary forensic artifacts from different filters for enhanced manipulation detection and localization.

Findings

MMFusion outperforms state-of-the-art models on multiple datasets.

Combining filters via early fusion improves detection accuracy.

Analysis of filter contributions enhances understanding of manipulation types.

Abstract

Recent image manipulation localization and detection techniques typically leverage forensic artifacts and traces that are produced by a noise-sensitive filter, such as SRM or Bayar convolution. In this paper, we showcase that different filters commonly used in such approaches excel at unveiling different types of manipulations and provide complementary forensic traces. Thus, we explore ways of combining the outputs of such filters to leverage the complementary nature of the produced artifacts for performing image manipulation localization and detection (IMLD). We assess two distinct combination methods: one that produces independent features from each forensic filter and then fuses them (this is referred to as late fusion) and one that performs early mixing of different modal outputs and produces combined features (this is referred to as early fusion). We use the latter as a feature encoding mechanism, accompanied by a new decoding mechanism that encompasses feature re-weighting, for formulating the proposed MMFusion architecture. We demonstrate that MMFusion achieves competitive performance for both image manipulation localization and detection, outperforming state-of-the-art models across several image and video datasets. We also investigate further the contribution of each forensic filter within MMFusion for addressing different types of manipulations, building on recent AI explainability measures.