Unseen objects, few-reference and open-vocabulary methods, category-level pose ambiguity, tracking, industrial data, multi-instance registration, and articulated digital twins.

Research trends, main themes, statistics, Highlight recommendations, and summaries of 17 papers.

Unseen-object generalization, category-level and articulated-part methods, multiview tracking, datasets, and efficient deployment.

Unseen objects and few references, structured category-level representations, environmental robustness, reliability, and precision measurement.

Based on the preliminary accepted-paper list: category-level flow / memory, arbitrary references, challenging industrial scenes, multi-object tracking, and hand–object interaction. Updated as the official proceedings become available.

Large-vocabulary category-level pose, foundation-model matching, unseen-object generalization, articulated and flexible objects, dynamic benchmarks, and CAD alignment.

Zero-shot category-level pose, articulated parts from a single RGB image, freely moving objects, and real-time hand–object interaction.

Few-reference unseen-object generalization, category-level articulated parts, hand–object interaction, monocular shape and pose, and model security.

Few-shot 6D pose, neural object representations, sparse-view alignment, physics-based tracking, and hand–object interaction states.

Novel-object 6D pose, alignment from unknown viewpoints, dynamic object tracking, and template-free hand–object interaction.

Few-shot novel-object pose, CAD alignment, partial-observation canonicalization, hand–object reconstruction, and articulated-object digital twins.

Generalization with limited priors, NeRF / NGP object representations, challenging industrial scenes, and video- and event-camera-based tracking.

Industrial zero-shot 6D pose, category-level and articulated objects, multiple shape–pose hypotheses, and visual–tactile in-hand pose.

6D / 9D pose with limited priors, event-camera tracking, visual–tactile fusion, pose uncertainty, and benchmarks for real robotic scenes.

Four themes: industrial symmetric objects, category-level geometric generalization, tool pose with limited priors, and dynamic reconstruction systems.

Category-level canonical representations, retrieval and efficient estimation, dynamic tracking, challenging materials, and surgical robotic instrument pose.

Unseen objects and occlusion, multiview depth and stereo, real-time spacecraft pose, rotation-only inference, and per-image pose-distribution evaluation.

Single-reference relative rotation, downstream effects of reconstructed models on pose, shape bias for challenging materials, and robotic rearrangement with generated goals.

Keypoint and training-data design, diffusion-augmented synthetic data, active markers for event cameras, and robotic perception under controlled lighting.