DynStream: Online Streaming 4D Gaussian Reconstruction of Dynamic Worlds from Unposed Video

Key points

• Online reconstruction of dynamic 4D scenes from long, unposed streaming videos requires both continuous processing and photorealistic rendering, which existing methods struggle to achieve simultaneously.1
• Existing feed-forward Gaussian methods are restricted to offline processing, whereas online point-cloud approaches struggle to maintain dense geometry and high-fidelity rendering.1
• Given a continuous video stream, DynStream reconstructs the scene within local temporal windows and incrementally aligns and fuses these local reconstructions into a globally consistent scene, enabling online 4D reconstruction without per-scene optimization.1
• Experiments demonstrate that DynStream enables high-fidelity online dynamic reconstruction and rendering from long video streams, achieving state-of-the-art performance across diverse dynamic indoor and outdoor scenes.1

Sources (1)

• [1]DynStream: Online Streaming 4D Gaussian Reconstruction of Dynamic Worlds from Unposed Video
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 09:15 AM

  We present DynStream, a framework for streaming 4D Gaussian reconstruction from long, unposed videos.

  Online reconstruction of dynamic 4D scenes from long, unposed streaming videos requires both continuous processing and photorealistic rendering, which existing methods struggle to achieve simultaneously.

Extractive summary: sentences quoted from the sources.