Bi-Erasing: A Bidirectional Framework for Concept Removal in Diffusion Models

TL;DR

Bi-Erasing introduces a bidirectional framework that simultaneously suppresses harmful concepts and promotes safe alternatives in diffusion models, improving concept removal while maintaining image quality.

Contribution

It proposes a novel bidirectional approach with dual image branches for balanced concept erasure and safety enhancement in diffusion models.

Findings

Outperforms baseline methods in concept removal effectiveness.

Maintains higher visual fidelity during concept erasure.

Effectively balances safety and image quality.

Abstract

Concept erasure, which fine-tunes diffusion models to remove undesired or harmful visual concepts, has become a mainstream approach to mitigating unsafe or illegal image generation in text-to-image models.However, existing removal methods typically adopt a unidirectional erasure strategy by either suppressing the target concept or reinforcing safe alternatives, making it difficult to achieve a balanced trade-off between concept removal and generation quality. To address this limitation, we propose a novel Bidirectional Image-Guided Concept Erasure (Bi-Erasing) framework that performs concept suppression and safety enhancement simultaneously. Specifically, based on the joint representation of text prompts and corresponding images, Bi-Erasing introduces two decoupled image branches: a negative branch responsible for suppressing harmful semantics and a positive branch providing visual guidance for safe alternatives. By jointly optimizing these complementary directions, our approach achieves a balance between erasure efficacy and generation usability. In addition, we apply mask-based filtering to the image branches to prevent interference from irrelevant content during the erasure process. Across extensive experiment evaluations, the proposed Bi-Erasing outperforms baseline methods in balancing concept removal effectiveness and visual fidelity.