SpaceDG: Benchmarking Spatial Intelligence under Visual Degradation

TL;DR

This paper introduces SpaceDG, a large-scale dataset and benchmark for evaluating and improving the robustness of multimodal large language models in spatial reasoning under various visual degradations.

Contribution

It presents the first degradation-aware spatial understanding dataset and benchmark, along with findings on model robustness and the benefits of degradation-aware finetuning.

Findings

Visual degradations significantly impair spatial reasoning in MLLMs.

Finetuning on SpaceDG improves robustness and can surpass human performance on degraded images.

The dataset contains about 1 million QA pairs across nine degradation types.

Abstract

Multimodal Large Language Models (MLLMs) have made rapid progress in spatial intelligence, yet existing spatial reasoning benchmarks largely assume pristine visual inputs and overlook the degradations that commonly occur in real-world deployment, such as motion blur, low light, adverse weather, lens distortion, and compression artifacts. This raises a fundamental question: how robust is the spatial intelligence of current MLLMs when visual observations are imperfect? To answer this question, we introduce SpaceDG, the first large-scale dataset for degradation-aware spatial understanding. It is constructed with a physically grounded degradation synthesis engine that embeds degradation formation process into 3D Gaussian Splatting (3DGS) rendering, enabling realistic simulation of nine degradation types. The resulting dataset contains approximately 1M QA pairs from nearly 1,000 indoor scenes. We further introduce SpaceDG-Bench, an human-verified benchmark with 1,102 questions spanning 11 reasoning categories and 9 visual degradation types, yielding over 10K VQA instances. Evaluating 25 open- and closed-source MLLMs reveals that visual degradations consistently and substantially impair spatial reasoning, exposing a critical robustness gap. Finally, we show that finetuning on SpaceDG markedly improves degradation robustness and can even surpass human performance under degraded conditions without any performance drop on clean images, highlighting the promise of degradation-aware training for robust spatial intelligence.