Region-Constrained Group Relative Policy Optimization for Flow-Based Image Editing

• URL: http://arxiv.org/abs/2604.09386v1
• Date: Fri, 10 Apr 2026 14:58:15 GMT
• Title: Region-Constrained Group Relative Policy Optimization for Flow-Based Image Editing
• Abstract summary: We propose RC-GRPO-Editing, a region-constrained GRPO post-training framework for flow-based image editing under deterministic ODE sampling.<n>It suppresses background-induced nuisance variance to enable cleaner localized credit assignment, improving region instruction adherence and preserving non-target content.<n> Experiments on CompBench show consistent improvements in editing region instruction adherence and non-target preservation.
• Score: 2.096755686662369
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Instruction-guided image editing requires balancing target modification with non-target preservation. Recently, flow-based models have emerged as a strong and increasingly adopted backbone for instruction-guided image editing, thanks to their high fidelity and efficient deterministic ODE sampling. Building on this foundation, GRPO-based reward-driven post-training has been explored to directly optimize editing-specific rewards, improving instruction following and editing consistency. However, existing methods often suffer from noisy credit assignment: global exploration also perturbs non-target regions, inflating within-group reward variance and yielding noisy GRPO advantages. To address this, we propose RC-GRPO-Editing, a region-constrained GRPO post-training framework for flow-based image editing under deterministic ODE sampling. It suppresses background-induced nuisance variance to enable cleaner localized credit assignment, improving editing region instruction adherence while preserving non-target content. Concretely, we localize exploration via region-decoupled initial noise perturbations to reduce background-induced reward variance and stabilize GRPO advantages, and introduce an attention concentration reward that aligns cross-attention with the intended editing region throughout the rollout, reducing unintended changes in non-target regions. Experiments on CompBench show consistent improvements in editing region instruction adherence and non-target preservation.

Related papers

• SR-Edit: Region-Aware Image Editing via Self-Refinement [30.465634862533516]
  SR-Edit is an image editing framework that overcomes issues via iterative self-refinement.<n>We show that SR-Edit achieves superior preservation and overall image quality compared to existing editing techniques.
  arXiv  Detail & Related papers  (2026-09-02T12:08:23Z)
• SpatialFlow-GRPO: Where Spatial Credit Drives Image Editing [29.67956120826966]
  We propose SpatialFlow-GRPO, a training framework that introduces spatially fine-grained reward feedback.<n>The framework converts region-aware rewards into semantic-region-level optimization signals and aligns region advantages with the corresponding latent positions during policy updates.<n>The results show that SpatialFlow-GRPO converts local feedback into spatially aligned update signals and improves editing quality.
  arXiv  Detail & Related papers  (2026-06-25T10:58:25Z)
• Edit-GRPO: A Locality-Preserving Policy Optimization Framework for Image Editing [95.29911956874389]
  We propose Edit-GRPO, preserving Locality while optimizing image editing.<n>By assigning region-specific optimization signals to edit and non-edit areas, Edit-GRPO aligns policy updates with the spatial structure of editing tasks.<n>This design effectively suppresses common artifacts such as context distortion and boundary inconsistency.
  arXiv  Detail & Related papers  (2026-05-16T12:05:39Z)
• Drag within Prior Distribution: Text-Conditioned Point-Based Image Editing within Distribution Constraints [7.503616785263929]
  Diffusion-based point editing can manipulate image semantics and fine details by applying localized perturbations on the manifold of noise latent perturbations.<n>Traditional point-based editing relies on pairs of handle and target points to define motion trajectories.<n>We introduce a CLIP-based model to evaluate and guide intermediate editing steps, ensuring that the generated results remain both semantically aligned.
  arXiv  Detail & Related papers  (2026-05-13T11:05:31Z)
• RefineAnything: Multimodal Region-Specific Refinement for Perfect Local Details [41.48256151708512]
  We introduce region-specific image refinement as a dedicated problem setting.<n>The goal is to restore fine-grained details while keeping all non-edited pixels strictly unchanged.<n>Existing instruction-driven editing models emphasize coarse-grained semantic edits.
  arXiv  Detail & Related papers  (2026-04-08T09:32:15Z)
• Adaptive Auxiliary Prompt Blending for Target-Faithful Diffusion Generation [6.826122099204317]
  We introduce Adaptive Auxiliary Prompt Blending (AAPB) to stabilize the diffusion process in low-density regions.<n>AAPB provides semantic support in rare concept generation and structural support in image editing.<n>We empirically show consistent improvements on the RareBench and FlowEdit datasets.
  arXiv  Detail & Related papers  (2026-03-19T17:12:03Z)
• A Turn Toward Better Alignment: Few-Shot Generative Adaptation with Equivariant Feature Rotation [67.2019317630466]
  Few-shot image generation aims to effectively adapt a source generative model to a target domain using very few training images.<n>We propose Equivariant Feature Rotation (EFR), a novel adaptation strategy that aligns source and target domains at two complementary levels.<n>Our method significantly enhances the generative performance within the targeted domain.
  arXiv  Detail & Related papers  (2025-12-24T13:48:22Z)
• EditMGT: Unleashing Potentials of Masked Generative Transformers in Image Editing [84.7089707244905]
  Masked Generative Transformers (MGTs) exhibit a localized decoding paradigm that endows them with the inherent capacity to preserve non-relevant regions during the editing process.<n>We introduce the first MGT-based image editing framework, termed EditMGT.<n>We demonstrate that EditMGT's cross-attention maps provide informative localization signals for localizing edit-relevant regions.<n>We also introduce region-hold sampling, which restricts token flipping within low-attention areas to suppress spurious edits.
  arXiv  Detail & Related papers  (2025-12-12T16:51:19Z)
• GRPO-Guard: Mitigating Implicit Over-Optimization in Flow Matching via Regulated Clipping [63.33669214116784]
  GRPO-Guard is a simple yet effective enhancement to existing GRPO frameworks.<n>It restores a balanced and step-consistent importance ratio, ensuring that PPO clipping properly constrains harmful updates.<n>It substantially mitigates implicit over-optimization without relying on heavy KL regularization.
  arXiv  Detail & Related papers  (2025-10-25T14:51:17Z)
• Follow-Your-Shape: Shape-Aware Image Editing via Trajectory-Guided Region Control [52.87568958372421]
  Follow-Your-Shape is a training-free and mask-free framework that supports precise and controllable editing of object shapes.<n>We compute a Trajectory Divergence Map (TDM) by comparing token-wise velocity differences between the inversion and denoising paths.<n>Our method achieves superior editability and visual fidelity, particularly in tasks requiring large-scale shape replacement.
  arXiv  Detail & Related papers  (2025-08-11T16:10:00Z)
• LORE: Latent Optimization for Precise Semantic Control in Rectified Flow-based Image Editing [0.276240219662896]
  We introduce LORE, a training-free and efficient image editing method.<n>LORE directly optimize the inverted noise, addressing the core limitations in generalization and controllability of existing approaches.<n> Experimental results show that LORE significantly outperforms strong baselines in terms of semantic alignment, image quality, and background fidelity.
  arXiv  Detail & Related papers  (2025-08-05T06:45:04Z)
• Uniform Attention Maps: Boosting Image Fidelity in Reconstruction and Editing [66.48853049746123]
  We analyze reconstruction from a structural perspective and propose a novel approach that replaces traditional cross-attention with uniform attention maps.<n>Our method effectively minimizes distortions caused by varying text conditions during noise prediction.<n> Experimental results demonstrate that our approach not only excels in achieving high-fidelity image reconstruction but also performs robustly in real image composition and editing scenarios.
  arXiv  Detail & Related papers  (2024-11-29T12:11:28Z)
• LIME: Localized Image Editing via Attention Regularization in Diffusion Models [69.33072075580483]
  This paper introduces LIME for localized image editing in diffusion models.<n>LIME does not require user-specified regions of interest (RoI) or additional text input, but rather employs features from pre-trained methods and a straightforward clustering method to obtain precise editing mask.<n>We propose a novel cross-attention regularization technique that penalizes unrelated cross-attention scores in the RoI during the denoising steps, ensuring localized edits.
  arXiv  Detail & Related papers  (2023-12-14T18:59:59Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.