DynaTok: Token-Based 4D Reconstruction from Partial Point Clouds
This work addresses the challenging problem of 4D reconstruction from incomplete, unordered point cloud sequences, which is relevant for applications like autonomous driving and robotics.
DynaTok proposes a point-based framework for 4D reconstruction from partial point cloud sequences without requiring correspondences or images, achieving improved reconstruction quality and temporal coherence on object- and scene-level benchmarks.
We address 4D reconstruction from partial point cloud sequences, where depth-sensor observations are incomplete, unordered, and lack explicit temporal correspondences. This geometry-only setting is challenging due to missing observations and ambiguous dynamics. While recent progress has largely relied on image-based methods, existing point-based approaches typically focus on single objects, assume relatively complete inputs, or require explicit correspondences. To address these limitations, we propose DynaTok, a point-based framework for correspondence-free 4D reconstruction from partial point cloud sequences without images. DynaTok encodes frames into compact latent tokens, aggregates incomplete observations over time with a Transformer-based spatiotemporal encoder, and decouples geometry and motion through residual tokens in a unified model. A flow-matching decoder then reconstructs complete, temporally consistent 4D point-cloud sequences conditioned on the latent tokens. Experiments on object- and scene-level benchmarks demonstrate improved reconstruction quality and temporal coherence from partial point cloud observations. Project page: https://wrchen530.github.io/dynatok/.