Spatial-TTT: Streaming Visual-based Spatial Intelligence with Test-Time Training
2026-03-12 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionMachine Learning
AI summaryⓘ
The authors developed a method called Spatial-TTT to help computers understand and remember the layout of places from long videos. Instead of just looking at short bits, their approach keeps updating and organizing spatial info as more video frames come in, using a special design that focuses on key spatial details and how scenes change over time. They also created a new dataset with detailed 3D descriptions to teach the model better. Their tests show this method improves the ability to understand large, complex spaces in videos.
spatial intelligencetest-time training (TTT)3D spatiotemporal convolutionsliding-window attentionfast weightslong-horizon video processingspatial-predictive mechanismvideo spatial benchmarks3D spatial descriptionshybrid neural architecture
Authors
Fangfu Liu, Diankun Wu, Jiawei Chi, Yimo Cai, Yi-Hsin Hung, Xumin Yu, Hao Li, Han Hu, Yongming Rao, Yueqi Duan
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.