MemPose: Category-level Object Pose Estimation with Memory
2026-07-06 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors propose MemPose, a new approach to estimating the orientation of objects in a category by using a special memory system. Unlike previous methods that use fixed shapes or static parameters, MemPose keeps a memory of different object structures it has seen before and updates this memory over time. This helps the system better understand and predict poses for a wide variety of objects within the same category. They tested their method on four different datasets and found it works better than earlier techniques.
category-level pose estimationobject posememory-augmented networksgeometric memorystructural representationsparametric modelsbenchmark datasetsREAL275CAMERA25
Authors
Xiao Lin, Minghao Zhu, Yun Peng, Liuyi Wang, Qiyi Wang, Chengju Liu, Qijun Chen
Abstract
In the pursuit of robust and generalizable category-level object pose estimation, most existing methods adopt parametric formulations that learn effective representations from data, yet they primarily encode category-level patterns into fixed shape priors or static parameter weights, which limits their scalability to highly diverse instances. In this paper, we rethink category-level pose estimation from a memory-centric perspective and present MemPose, a memory-augmented framework that explicitly incorporates category-level geometric memory into the pose estimation pipeline. We introduce an external memory buffer that stores and dynamically updates structural representations from previously observed instances, enabling the model to leverage accumulated experience to support current perception. Extensive experiments on four challenging benchmarks (REAL275, CAMERA25, Housecat6D and Wild6D) demonstrate the superiority of our proposed method over previous state-of-the-art approaches.