Error-Driven Scene Editing for 3D Grounding in Large Language Models

TL;DR

This paper introduces DEER-3D, an error-driven scene editing framework that improves 3D grounding in large language models by generating targeted visual counterfactuals through minimal, predicate-specific scene modifications.

Contribution

It presents a novel error-driven approach for 3D scene editing that enhances grounding accuracy without extensive 3D data collection or scene reconstruction.

Findings

Consistent performance improvements across multiple 3D grounding benchmarks.

Effective diagnosis and correction of predicate-level grounding errors.

Enhanced spatial understanding in 3D LLMs through targeted scene edits.

Abstract

Despite recent progress in 3D-LLMs, they remain limited in accurately grounding language to visual and spatial elements in 3D environments. This limitation stems in part from training data that focuses on language reasoning rather than spatial understanding due to scarce 3D resources, leaving inherent grounding biases unresolved. To address this, we propose 3D scene editing as a key mechanism to generate precise visual counterfactuals that mitigate these biases through fine-grained spatial manipulation, without requiring costly scene reconstruction or large-scale 3D data collection. Furthermore, to make these edits targeted and directly address the specific weaknesses of the model, we introduce DEER-3D, an error-driven framework following a structured "Decompose, Diagnostic Evaluation, Edit, and Re-train" workflow, rather than broadly or randomly augmenting data as in conventional approaches. Specifically, upon identifying a grounding failure of the 3D-LLM, our framework first diagnoses the exact predicate-level error (e.g., attribute or spatial relation). It then executes minimal, predicate-aligned 3D scene edits, such as recoloring or repositioning, to produce targeted counterfactual supervision for iterative model fine-tuning, significantly enhancing grounding accuracy. We evaluate our editing pipeline across multiple benchmarks for 3D grounding and scene understanding tasks, consistently demonstrating improvements across all evaluated datasets through iterative refinement. DEER-3D underscores the effectiveness of targeted, error-driven scene editing in bridging linguistic reasoning capabilities with spatial grounding in 3D LLMs.