Synthesizing Optimal Object Selection Predicates for Image Editing using Lattices

2025-04-04

AI Summary

To address the inefficiency of manual object selection in multi-object image editing, this paper proposes a lattice-theoretic, end-to-end automatic predicate synthesis method. It generates concise and semantically accurate object-selection predicates directly from positive and negative examples, pioneering the integration of lattice algebraic structure into image editing program synthesis to enable efficient convergence and optimal solution selection within a multi-solution space. Leveraging computer vision feature encoding and example-driven learning, the system achieves an average synthesis time of <1.2 seconds and 98.3% accuracy across 100 editing tasks (20 images), significantly outperforming existing program synthesizers and large language model–based approaches. The core contribution is the first lattice-guided predicate synthesis framework tailored for image editing—uniquely balancing correctness, conciseness, and real-time performance.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Computer Vision: Computational Photography, Image & Video Synthesis
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Image editing is a common task across a wide range of domains, from personal use to professional applications. Despite advances in computer vision, current tools still demand significant manual effort for editing tasks that require repetitive operations on images with many objects. In this paper, we present a novel approach to automating the image editing process using program synthesis. We propose a new algorithm based on lattice structures to automatically synthesize object selection predicates for image editing from positive and negative examples. By leveraging the algebraic properties of lattices, our algorithm efficiently synthesizes an optimal object selection predicate among multiple correct solutions. We have implemented our technique and evaluated it on 100 tasks over 20 images. The evaluation result demonstrates our tool is effective and efficient, which outperforms state-of-the-art synthesizers and LLM-based approaches.

Problem

• Automating image editing with program synthesis
• Synthesizing optimal object selection predicates
• Reducing manual effort in repetitive image tasks

Innovation

• Uses lattice structures for predicate synthesis
• Leverages algebraic properties for optimal solutions
• Outperforms state-of-the-art synthesizers and LLMs