AI summaryⓘ
The authors address the problem of estimating the position and orientation (pose) of objects within a category, even when the objects look different or are partially hidden. They note that current methods rely heavily on feature learning but often produce unstable results when objects change shape or are occluded. To improve this, the authors introduce TriCons-Pose, a method that detects stable keypoints consistently across views and uses triangle-based geometric information to better represent object poses. Their method includes a keypoint detector that ensures structural consistency and an aggregator that enhances pose features using geometric cues. Experiments on several benchmark datasets show that their approach is effective.
Category-level object pose estimationKeypoint detectionGeometric consistencyPose invarianceCanonical mappingCross-view consistencyAttention mechanismTriangle-based descriptorsOcclusion handling3D object recognition
Authors
Zuzhi Yang, Shuai Wang, Mounir Kaaniche, Ziwei Li, Zhiming Cheng, Zhidong Zhao, Chenggang Yan
Abstract
Category-level object pose estimation is a crucial yet challenging task in both academia and industry, and has achieved remarkable success by leveraging keypoint-based correspondence paradigms. However, most existing methods increasingly rely on stronger feature learning while overlooking whether the established correspondences are geometrically stable across diverse perturbations. This often results in fragile pose recovery under intra-class shape variations and occlusions. To tackle this challenge, we develop a novel Triangle-Invariant Geometric Consistency Learning for Category-Level Object Pose Estimation (TriCons-Pose) to anchor stable keypoints and aggregate pose-invariant cues, yielding reliable canonical mapping and accurate pose estimation. Specifically, a Structure-Consistent Keypoint Detector (SCKD) is designed to identify robust keypoints by enforcing cross-view structural consistency via normalized pairwise distance matching. Moreover, we propose a Pose-Invariant Geometric Aggregator (PIGA) to augment keypoint representations by injecting triangle-based pose-invariant descriptors into a local-to-global attention mechanism. The proposed framework is optimized using standard objective functions while incorporating an additional geometry consistency loss. Extensive experiments on REAL275, CAMERA25, and HouseCat6D datasets demonstrate the effectiveness of the proposed approach.