Multi-Tool Image Editing Attribution in Facial Forgery

Authors: Sheng Liu, Qiang Sheng, Danding Wang, Yu Li, Chenming Zhou, Juan Cao

Published: 2026-09-02 15:51:26+00:00

Comment: Accepted to ACM Multimedia 2026 (MM 2026)

AI Summary

This paper introduces Multi-Tool Image Editing Attribution (MIEA) to identify multiple editing tools used in a facial image, addressing the limitations of single-tool attribution methods. They construct a new dataset, MultiEdit, with over 500k edited facial images covering six tool types. The authors also propose DPEC, a method that captures locality-aware editing traces from both spatial and frequency domains, enhanced by an error-based curriculum learning strategy.

Abstract

As generative AI tools become increasingly powerful and easy to use, people can easily edit portrait images with a prompt, necessitating the task of image editing attribution, which predicts the involved editing tools from the given image. Existing attribution methods hold the single-tool assumption and can only attribute a specific editing tool, but struggle to handle the more complex and increasingly common multi-tool editing scenarios, where artifacts left by different editing tools are composite and overlapped. To address this gap, we explore Multi-Tool Image Editing Attribution (MIEA), which aims to identify multiple editing tools involved in a multi-tool edited facial image. To simulate the real-life editing operations on facial images, we then construct a new dataset, MultiEdit, which contains 500k+ edited facial images and covers six types of editing tools that support face swapping (Deepfake) and various facial enhancements. Inspired by the findings from data analysis, we design DPEC, a multi-tool attribution method that can capture distinguishable, locality-aware editing tool traces from both spatial and frequency domains with the support of an error-based curriculum learning strategy. Experiments show \\Method\\ outperforms nine methods for facial images edited in at most five steps.

Key findings

DPEC significantly outperforms nine baseline methods for facial images edited with up to five steps, achieving a mean strict accuracy of 92.93% and a mean toolwise accuracy of 98.56%. The study also reveals that multi-tool editing makes individual tool detection harder, and the proposed PFE and ECL modules are crucial for DPEC's performance. DPEC also achieves competitive performance in traditional binary deepfake detection and shows robustness against image compression.

Approach

The authors formulate multi-tool image editing attribution as a multi-label classification task. Their proposed DPEC method utilizes a Patchwise Feature Enhancement (PFE) module to extract distinguishable, locality-aware features from both spatial and frequency domains, and an Error-based Curriculum Learning (ECL) strategy to progressively train the model from easy to hard samples based on error feedback.

Datasets

MultiEdit (a new dataset constructed by the authors, containing 500k+ edited facial images), DEFACTO-Face, FF++, Celeb-DF

Model(s)

DPEC (proposed), ResNet18, ResNet50, Xception, SwinT, F3-Net, Qwen3-VL, Qwen3-VL-CLS, NPR, FatFormer

Author countries

China

← Previous