Category Agnostic Pose Estimation

Category-Agnostic Pose Estimation (CAPE) is a subtask in the field of computer vision that aims to develop a model capable of detecting the poses of objects of any category using only a few samples with keypoint definitions. The core objective of this task is to achieve efficient and accurate pose estimation for unknown categories, thereby enhancing the model's generalization and adaptability. CAPE holds significant value in applications such as robotic navigation, augmented reality, and automated inspection, as it can substantially improve the flexibility and robustness of systems.