The Flow has not summarised this story yet — read it at arXiv Computer Vision.
TAPVid-MV is a new benchmark for tracking any point in 3D across multiple synchronized camera views. It comprises 284 sequences, 1,142 calibrated camera streams, and 109,769 point tracks, covering indoor and outdoor domains and derived from various modalities such as depth, LiDAR, SLAM, and simulation. The dataset is visually verified, and evaluation shows that current multi‑view trackers do not consistently outperform monocular trackers, highlighting geometry recovery as a key bottleneck.
Point2Pose is a model‑free method for causal 6D pose tracking of multiple rigid objects using monocular RGB‑D video. It starts from sparse image points and employs a 2D point tracker to maintain long‑range correspondences, allowing instant recovery after complete occlusion. The system also incrementally builds an online Truncated Signed Distance Function (TSDF) representation of the tracked objects and introduces a new multi‑object tracking dataset with motion‑capture ground truth.
Track2Art is a motion‑centric framework that recovers articulated object models from RGB‑D interaction videos by lifting 2D point tracks into 3D trajectories. It groups these trajectories into rigid‑part hypotheses and uses learned‑analytic reasoning to infer directed kinematic relations, joint types, and joint geometry. On the PartNet‑Mobility benchmark, it achieves 0.695 Point IoU and 0.410 end‑to‑end J@20 without requiring ground‑truth part counts or test‑time optimization.
The paper introduces MoMa-SG, a framework that constructs semantic‑kinematic 3D scene graphs for articulated scenes using RGB‑D sequences. It segments interactions, tracks points robustly, estimates joint parameters in a single optimization, and associates objects with articulations to detect contained items. The authors also present the Arti4D‑Semantic dataset with 62 real‑world sequences and demonstrate the system’s effectiveness on a quadruped and a mobile manipulator.
MoSE3 is a feed‑forward model that predicts dense SE(3) motion—full 6‑DoF rigid transforms—at every pixel from monocular RGB video, providing rotation, translation, and grouping information simultaneously. It achieves this by jointly learning 3D point tracks and rigidity embeddings, then differentiably fitting transforms within soft rigid clusters. The authors also release Art‑Kubric, a large synthetic dataset with dense SE(3) and rigidity labels, and demonstrate state‑of‑the‑art performance on both rigid and articulated benchmarks, with strong generalization to real‑world videos.
arXiv:2609.27675v2 Announce Type: replace Abstract: Understanding articulated objects is fundamental for robotic interaction, requiring accurate rigid-part discovery and the recovery of their kinemat...