Aug 16, 2018 · We propose to directly regress a pose vector from raw point cloud segments using a convolutional neural network.

We formulate the problem of estimating the rotation of a rigid object as regression from a point cloud segment to the axis-angle representation of the rotation.

Jan 23, 2019 · We formulate the problem of estimating the rotation of a rigid object as regression from a point cloud segment to the axis-angle representation ...

Experimental results show that our method achieves competitive performance compared to a state-of-the-art method, while also showing more robustness against ...

We propose to directly regress a pose vector from point cloud segments using a convolutional neural network. Experimental results show that our method achieves ...

Occlusion Resistant Object Rotation Regression from Point Cloud Segments ... Rotation estimation of known rigid objects is important for robotic applications such ...

Occlusion resistant object rotation regression from point cloud segments. G Gao, M Lauri, J Zhang, S Frintrop. Proceedings of the European Conference on ...

(ECCV2018) Occlusion Resistant Object Rotation Regression from Point Cloud Segments [paper link]. DeepIM(ECCV2018) DeepIM: Deep Iterative Matching for 6D ...

Frintrop, “Occlusion resistant object rotation regression from point cloud segments,” in ECCV 4th. International Workshop on Recovering 6D Object Pose, 2018.

In this study, we present a novel approach that leverages deep neural networks to learn a model capable of filling gaps in urban scenes that are obscured by ...