Search papers, labs, and topics across Lattice.
This paper introduces the Global-Spatial-Temporal Benchmark (GST-Bench), a novel video question-answering benchmark designed to evaluate global spatial intelligence in visual understanding. By analyzing 22 state-of-the-art vision-language models (VLMs) against human performance, the authors reveal a significant gap in spatial reasoning capabilities, with the best model scoring only 42.68 compared to the human score of 79.08. The study highlights that while models excel in local spatial tasks, they struggle to integrate long-horizon observations into a coherent global scene representation, prompting the development of GST-Train to aid future research in this area.
VLMs are failing to achieve human-level global spatial awareness, scoring only 42.68 on a new benchmark compared to 79.08 for humans.
Spatial intelligence is fundamental to embodied agents, yet existing benchmarks focus on local spatial perception from single or few viewpoints, overlooking global spatial awareness over continuous, long-horizon visual streams. To address this limitation, we introduce the Global-Spatial-Temporal Benchmark (GST-Bench), a VQA benchmark for global spatial intelligence in video understanding, comprising human-verified questions derived from 6,790 minutes of synthetically generated video. It requires models to perform accurate spatial inference from novel viewpoints unseen in the input video and to map egocentric observations onto global top-down images. A comprehensive evaluation of 22 state-of-the-art VLMs exposes a striking gap between models and humans: the strongest zero-shot model attains only 42.68, far below the human score of 79.08. To probe the cause of this gap, we construct GST-Bench-Local and find that models, despite strong local spatial understanding under the same task formulation, still fail to consolidate long-horizon observations into a globally consistent scene representation. We further provide GST-Train, a dataset for global spatial reasoning, as a complementary resource to facilitate future research on this challenge.
