SEED: A Benchmark Dataset for Sequential Facial Attribute Editing with Diffusion Models

TL;DR

SEED is a large-scale dataset of sequential facial edits created with diffusion models, enabling research on progressive attribute changes, edit tracking, and robustness in facial image manipulation.

Contribution

The paper introduces SEED, a comprehensive dataset of over 90,000 images with sequential edits, and proposes FAITH, a frequency-aware transformer model for tracking subtle attribute changes.

Findings

FAITH outperforms baseline models in sequential edit sensitivity

SEED reveals challenges in robustness and attribution in progressive editing

Frequency cues improve detection of subtle attribute modifications

Abstract

Diffusion models have recently enabled precise and photorealistic facial editing across a wide range of semantic attributes. Beyond single-step modifications, a growing class of applications now demands the ability to analyze and track sequences of progressive edits, such as stepwise changes to hair, makeup, or accessories. However, sequential editing introduces significant challenges in edit attribution and detection robustness, further complicated by the lack of large-scale, finely annotated benchmarks tailored explicitly for this task. We introduce SEED, a large-scale Sequentially Edited facE Dataset constructed via state-of-the-art diffusion models. SEED contains over 90,000 facial images with one to four sequential attribute modifications, generated using diverse diffusion-based editing pipelines (LEdits, SDXL, SD3). Each image is annotated with detailed edit sequences, attribute masks, and prompts, facilitating research on sequential edit tracking, visual provenance analysis, and manipulation robustness assessment. To benchmark this task, we propose FAITH, a frequency-aware transformer-based model that incorporates high-frequency cues to enhance sensitivity to subtle sequential changes. Comprehensive experiments, including systematic comparisons of multiple frequency-domain methods, demonstrate the effectiveness of FAITH and the unique challenges posed by SEED. SEED offers a challenging and flexible resource for studying progressive diffusion-based edits at scale. Dataset and code will be publicly released at: https://github.com/Zeus1037/SEED.