InterCoG: Towards Spatially Precise Image Editing with Interleaved Chain-of-Grounding Reasoning

2026-03-02

AI Summary

This work addresses the challenge of fine-grained, spatially precise image editing for visually inconspicuous targets in complex multi-entity scenes. The authors propose InterCoG, a framework that leverages spatial reasoning from textual instructions to localize editing targets, generates bounding boxes and masks via visual grounding, and rewrites editing commands to clarify intended outcomes. A key innovation is the introduction of a text–vision interleaved chain-of-grounding reasoning mechanism, augmented by multimodal grounding reconstruction supervision and a reasoning alignment module, which together significantly enhance localization accuracy and model interpretability. To support this research, the authors also construct the GroundEdit-45K dataset and the GroundEdit-Bench evaluation benchmark. Experimental results demonstrate that InterCoG substantially outperforms existing methods on fine-grained image editing tasks.

Technology Category

• Natural Language Processing: Language Grounding & Multi-modal NLP
• Computer Vision: Visual Reasoning & Symbolic Representations
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Emerging unified editing models have demonstrated strong capabilities in general object editing tasks. However, it remains a significant challenge to perform fine-grained editing in complex multi-entity scenes, particularly those where targets are not visually salient and require spatial reasoning. To this end, we propose InterCoG, a novel text-vision Interleaved Chain-of-Grounding reasoning framework for fine-grained image editing in complex real-world scenes. The key insight of InterCoG is to first perform object position reasoning solely within text that includes spatial relation details to explicitly deduce the location and identity of the edited target. It then conducts visual grounding via highlighting the editing targets with generated bounding boxes and masks in pixel space, and finally rewrites the editing description to specify the intended outcomes. To further facilitate this paradigm, we propose two auxiliary training modules: multimodal grounding reconstruction supervision and multimodal grounding reasoning alignment to enforce spatial localization accuracy and reasoning interpretability, respectively. We also construct GroundEdit-45K, a dataset comprising 45K grounding-oriented editing samples with detailed reasoning annotations, and GroundEdit-Bench for grounding-aware editing evaluation. Extensive experiments substantiate the superiority of our approach in highly precise edits under spatially intricate and multi-entity scenes.

Problem

• fine-grained image editing
• spatial reasoning
• multi-entity scenes
• visual grounding
• spatially precise editing

Innovation

• Interleaved Chain-of-Grounding
• spatially precise image editing
• visual grounding
• multimodal reasoning
• fine-grained editing