PoseMatcher: One-shot 6D Object Pose Estimation by Deep Feature Matching

TL;DR

PoseMatcher is a novel one-shot 6D object pose estimation method that uses deep feature matching, a new attention layer, and a pruning strategy to outperform prior methods on standard datasets.

Contribution

It introduces PoseMatcher, a model-free one-shot pose estimator with a new training pipeline, IO-Layer attention, and a pruning strategy, advancing the state of the art.

Findings

Outperforms all prior one-shot methods on Linemod and YCB-V datasets.

Achieves results comparable to recent instance-level pose estimation methods.

Introduces a new attention layer and pruning strategy to improve efficiency and accuracy.

Abstract

Estimating the pose of an unseen object is the goal of the challenging one-shot pose estimation task. Previous methods have heavily relied on feature matching with great success. However, these methods are often inefficient and limited by their reliance on pre-trained models that have not be designed specifically for pose estimation. In this paper we propose PoseMatcher, an accurate model free one-shot object pose estimator that overcomes these limitations. We create a new training pipeline for object to image matching based on a three-view system: a query with a positive and negative templates. This simple yet effective approach emulates test time scenarios by cheaply constructing an approximation of the full object point cloud during training. To enable PoseMatcher to attend to distinct input modalities, an image and a pointcloud, we introduce IO-Layer, a new attention layer that efficiently accommodates self and cross attention between the inputs. Moreover, we propose a pruning strategy where we iteratively remove redundant regions of the target object to further reduce the complexity and noise of the network while maintaining accuracy. Finally we redesign commonly used pose refinement strategies, zoom and 2D offset refinements, and adapt them to the one-shot paradigm. We outperform all prior one-shot pose estimation methods on the Linemod and YCB-V datasets as well achieve results rivaling recent instance-level methods. The source code and models are available at https://github.com/PedroCastro/PoseMatcher.