Diffusion Counterfactual Generation with Semantic Abduction

TL;DR

This paper introduces a diffusion-based framework for counterfactual image generation that leverages semantic abduction and causal reasoning to improve identity preservation and control over generated images.

Contribution

It presents a novel integration of semantic representations into diffusion models using Pearlian causality, enabling high-level semantic control in counterfactual image editing.

Findings

First to incorporate high-level semantic identity preservation in diffusion counterfactuals

Demonstrates how semantic control enables trade-offs between causal fidelity and identity preservation

Introduces a general framework for diffusion-based causal mechanisms with semantic abduction.

Abstract

Counterfactual image generation presents significant challenges, including preserving identity, maintaining perceptual quality, and ensuring faithfulness to an underlying causal model. While existing auto-encoding frameworks admit semantic latent spaces which can be manipulated for causal control, they struggle with scalability and fidelity. Advancements in diffusion models present opportunities for improving counterfactual image editing, having demonstrated state-of-the-art visual quality, human-aligned perception and representation learning capabilities. Here, we present a suite of diffusion-based causal mechanisms, introducing the notions of spatial, semantic and dynamic abduction. We propose a general framework that integrates semantic representations into diffusion models through the lens of Pearlian causality to edit images via a counterfactual reasoning process. To our knowledge, this is the first work to consider high-level semantic identity preservation for diffusion counterfactuals and to demonstrate how semantic control enables principled trade-offs between faithful causal control and identity preservation.