ParallelEdits: Efficient Multi-object Image Editing

2024-06-03

AI Summary

Text-driven image editing suffers from challenges in concurrently modifying multiple objects or attributes, while serial processing leads to low efficiency and degraded output quality. To address this, we propose the first diffusion-based framework enabling parallel multi-target editing. Our core contributions include: (1) an attention allocation mechanism that dynamically decouples feature responses across distinct editing regions; (2) a multi-branch parallel architecture that jointly modifies multiple attributes in a single denoising step; and (3) joint prompt embedding with branch-wise feature disentanglement to enhance semantic alignment accuracy. We further introduce PIE-Bench++, a comprehensive benchmark for evaluating multi-target editing under complex real-world scenarios. Extensive experiments demonstrate that our method significantly outperforms existing serial approaches in editing consistency, visual fidelity, and inference efficiency.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Search and Optimization: Distributed Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Text-driven image synthesis has made significant advancements with the development of diffusion models, transforming how visual content is generated from text prompts. Despite these advances, text-driven image editing, a key area in computer graphics, faces unique challenges. A major challenge is making simultaneous edits across multiple objects or attributes. Applying these methods sequentially for multi-attribute edits increases computational demands and efficiency losses. In this paper, we address these challenges with significant contributions. Our main contribution is the development of ParallelEdits, a method that seamlessly manages simultaneous edits across multiple attributes. In contrast to previous approaches, ParallelEdits not only preserves the quality of single attribute edits but also significantly improves the performance of multitasking edits. This is achieved through innovative attention distribution mechanism and multi-branch design that operates across several processing heads. Additionally, we introduce the PIE-Bench++ dataset, an expansion of the original PIE-Bench dataset, to better support evaluating image-editing tasks involving multiple objects and attributes simultaneously. This dataset is a benchmark for evaluating text-driven image editing methods in multifaceted scenarios.

Problem

• Enables simultaneous multi-object editing in images
• Reduces computational costs of sequential attribute edits
• Introduces benchmark dataset for multi-attribute editing evaluation

Innovation

• ParallelEdits enables simultaneous multi-attribute image editing
• Uses attention distribution and multi-branch design
• Introduces PIE-Bench++ dataset for multi-object evaluation