ICPose

ICPose is an acronym that appears in several research contexts related to pose in computer vision and graphics. There is no universally accepted definition; the term is used differently by different groups, making it a disambiguation rather than a single concept.

In some publications, ICPose refers to an iterative method for aligning or reconstructing human poses from image data, drawing on ideas related to pose estimation and optimization. In other lines of work, it designates a dataset or benchmark for indoor pose annotation and evaluation, intended to test robustness to occlusion, lighting, and viewpoint changes. A third common usage describes a compact pose descriptor or representation that facilitates tracking of objects or people in video sequences.

Because of its varied usage, researchers typically clarify the exact definition of ICPose in their papers, including the data modalities, evaluation metrics, and application domain. The term may intersect with related concepts such as pose estimation, 3D reconstruction, motion capture, and the Iterative Closest Point (ICP) framework, but it is not a standardized term across the literature.

See also: Pose estimation, 3D reconstruction, motion capture, Iterative Closest Point.