RePainter: Empowering E-commerce Object Removal via Spatial-matting Reinforcement Learning

2025-10-08

AI Summary

To address visual quality degradation in e-commerce images caused by watermarks and promotional text overlays, this paper proposes the first reinforcement learning–driven image inpainting framework tailored for commercial scenarios. Methodologically, it integrates diffusion models with a spatial-attention modulation mechanism, introduces spatial matting trajectory optimization, and employs Grouped Relative Policy Optimization (GRPO). A composite reward function is designed to jointly optimize global structure, local detail fidelity, and semantic consistency—effectively mitigating artifacts and reward hacking. To support reproducible research, we introduce EcomPaint-100K, a large-scale e-commerce image dataset, and EcomPaint-Bench, a dedicated evaluation benchmark. Extensive experiments demonstrate that our method significantly outperforms existing state-of-the-art approaches on challenging tasks involving complex compositions and persistent text/logo removal, yielding more natural and semantically coherent results. The code and models are publicly released.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Learning to Search
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

In web data, product images are central to boosting user engagement and advertising efficacy on e-commerce platforms, yet the intrusive elements such as watermarks and promotional text remain major obstacles to delivering clear and appealing product visuals. Although diffusion-based inpainting methods have advanced, they still face challenges in commercial settings due to unreliable object removal and limited domain-specific adaptation. To tackle these challenges, we propose Repainter, a reinforcement learning framework that integrates spatial-matting trajectory refinement with Group Relative Policy Optimization (GRPO). Our approach modulates attention mechanisms to emphasize background context, generating higher-reward samples and reducing unwanted object insertion. We also introduce a composite reward mechanism that balances global, local, and semantic constraints, effectively reducing visual artifacts and reward hacking. Additionally, we contribute EcomPaint-100K, a high-quality, large-scale e-commerce inpainting dataset, and a standardized benchmark EcomPaint-Bench for fair evaluation. Extensive experiments demonstrate that Repainter significantly outperforms state-of-the-art methods, especially in challenging scenes with intricate compositions. We will release our code and weights upon acceptance.

Problem

• Removes intrusive watermarks and promotional text from e-commerce images
• Improves unreliable object removal and domain adaptation in commercial settings
• Reduces visual artifacts and unwanted object insertion in product visuals

Innovation

• Integrates spatial-matting trajectory refinement with GRPO
• Modulates attention mechanisms to emphasize background context
• Introduces composite reward mechanism balancing multiple constraints