OpenGPT-4o-Image: A Comprehensive Dataset for Advanced Image Generation and Editing

2025-09-29

AI Summary

Existing unified multimodal image generation and editing models are constrained by the scale, systematicity, and task complexity of training data. To address this, we introduce the first large-scale, structured, high-quality dataset covering 11 domains and 51 subtasks, comprising 80,000 instruction–image pairs. Our method features a novel hierarchical task taxonomy—first incorporating scientifically grounded image generation and concurrent multi-step instruction execution—and an automated data generation pipeline integrating GPT-4o with a curated structural resource pool to ensure both diversity and controllability. Evaluated on ImgEdit-Bench and GenEval benchmarks, our model achieves up to 18% improvement in editing performance and 13% in generation performance over prior baselines. These advances significantly enhance the generalization capability and real-world applicability of multimodal models in complex, heterogeneous scenarios.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Multi-modal Vision
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

The performance of unified multimodal models for image generation and editing is fundamentally constrained by the quality and comprehensiveness of their training data. While existing datasets have covered basic tasks like style transfer and simple object manipulation, they often lack the systematic structure and challenging scenarios required for real-world applications. To address this bottleneck, we introduce OpenGPT-4o-Image, a large-scale dataset constructed using a novel methodology that combines hierarchical task taxonomy with automated data generation. Our taxonomy not only includes fundamental capabilities such as text rendering and style control but also introduces highly practical yet challenging categories like scientific imagery for chemistry illustrations and complex instruction editing requiring simultaneous execution of multiple operations. Through an automated pipeline leveraging structured resource pools and GPT-4o, we generate 80k high-quality instruction-image pairs with controlled diversity, covering 11 major domains and 51 subtasks. Extensive experiments show that fine-tuning leading models on our dataset achieves significant performance gains across multiple benchmarks, with improvements of up to 18% on editing tasks (UniWorld-V1 on ImgEdit-Bench) and 13% on generation tasks (Harmon on GenEval). Our work demonstrates that systematic data construction is key to advancing multimodal AI capabilities.

Problem

• Addressing limitations in multimodal training data quality
• Systematically covering complex real-world image editing scenarios
• Enhancing image generation and editing through structured datasets

Innovation

• Hierarchical task taxonomy for systematic data structure
• Automated pipeline generating 80k instruction-image pairs
• Covers 11 domains with scientific imagery and complex editing