Paper 2603.12255
Spatial-TTT: Streaming Visual-based Spatial Intelligence with Test-Time Training
- Published
- Mar 2026
- Research lab
- Independent
- Citations
- 3
- GitHub
- 249 stars
01 In brief
Summary
Spatial-TTT is a framework for streaming visual-based spatial intelligence that uses test-time training (TTT) to maintain adaptive fast weights as a compact memory for accumulating 3D evidence from long-horizon videos.
It employs a hybrid architecture interleaving TTT layers with self-attention anchor layers at a 3:1 ratio, large-chunk updates, and sliding-window attention for efficiency.
A spatial-predictive mechanism uses depth-wise 3D spatiotemporal convolutions to capture geometric correspondence and temporal continuity.
A dense scene-description dataset (16K samples from ScanNet and ARKitScenes) provides supervision for learning fast-weight update dynamics, followed by fine-tuning on 3M spatial VQA samples.
On VSI-Bench, Spatial-TTT-2B achieves an average score of 64.4, outperforming proprietary and open-source baselines; on MindCube-Tiny it scores 76.2, and on VSI-SUPER-Count it significantly outperforms baselines.
Ablations show each component contributes, and the model scales linearly in memory and compute, unlike quadratic baselines.
02 From the paper
Abstract
Humans perceive and understand real-world spaces through a stream of visual observations. Therefore, the ability to streamingly maintain and update spatial evidence from potentially unbounded video streams is essential for spatial intelligence. The core challenge is not simply longer context windows but how spatial information is selected, organized, and retained over time. In this paper, we propose Spatial-TTT towards streaming visual-based spatial intelligence with test-time training (TTT), which adapts a subset of parameters (fast weights) to capture and organize spatial evidence over long-horizon scene videos. Specifically, we design a hybrid architecture and adopt large-chunk updates parallel with sliding-window attention for efficient spatial video processing. To further promote spatial awareness, we introduce a spatial-predictive mechanism applied to TTT layers with 3D spatiotemporal convolution, which encourages the model to capture geometric correspondence and temporal continuity across frames. Beyond architecture design, we construct a dataset with dense 3D spatial descriptions, which guides the model to update its fast weights to memorize and organize global 3D spatial signals in a structured manner. Extensive experiments demonstrate that Spatial-TTT improves long-horizon spatial understanding and achieves state-of-the-art performance on video spatial benchmarks. Project page: https://liuff19.github.io/Spatial-TTT.