LayoutRectifier: An Optimization-based Post-processing for Graphic Design Layout Generation

2025-08-14

AI Summary

Existing deep learning–based layout generation methods frequently exhibit structural flaws, including element misalignment, overlap, and semantically inconsistent containment relationships. To address these issues without retraining, this paper proposes a two-stage post-processing optimization framework. In the first stage, a discrete search guided by a professional grid system precisely corrects misalignments. In the second stage, a differentiable box containment function is introduced to jointly optimize positions and dimensions, thereby eliminating overlaps and improving containment plausibility. The method synergistically combines combinatorial search with gradient-based optimization, achieving high-fidelity corrections while minimizing deviation from the original layout. Experiments demonstrate significant improvements in structural compliance and design consistency across both content-agnostic and content-aware layout generation tasks. The framework is plug-and-play, effectively enhancing the output quality of mainstream generative models without architectural or training modifications.

Technology Category

• Search and Optimization: Mixed Discrete/Continuous Search
• Computer Vision: Learning & Optimization for CV
• Constraint Satisfaction and Optimization: Mixed Discrete/Continuous Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Recent deep learning methods can generate diverse graphic design layouts efficiently. However, these methods often create layouts with flaws, such as misalignment, unwanted overlaps, and unsatisfied containment. To tackle this issue, we propose an optimization-based method called LayoutRectifier, which gracefully rectifies auto-generated graphic design layouts to reduce these flaws while minimizing deviation from the generated layout. The core of our method is a two-stage optimization. First, we utilize grid systems, which professional designers commonly use to organize elements, to mitigate misalignments through discrete search. Second, we introduce a novel box containment function designed to adjust the positions and sizes of the layout elements, preventing unwanted overlapping and promoting desired containment. We evaluate our method on content-agnostic and content-aware layout generation tasks and achieve better-quality layouts that are more suitable for downstream graphic design tasks. Our method complements learning-based layout generation methods and does not require additional training.

Problem

• Rectifies flaws in auto-generated graphic design layouts
• Reduces misalignment and unwanted overlaps in layouts
• Optimizes layouts without requiring additional training

Innovation

• Optimization-based post-processing for layout rectification
• Two-stage optimization with grid systems
• Novel box containment function for adjustments