2018 · 30 citations · 15 references
ComputingComputer ScienceComputer VisionMachine LearningMachine VisionDeep LearningFeature DescriptionMulti-task LearningObject RecognitionPose EstimationRepresentation Learning3D Pose Estimation6D Pose EstimationConvolutional Neural NetworksDescriptor MatchingManifold LearningObject Pose EstimationTriplet LossTriplet Networks
In this work, we propose a method for object recognition and pose estimation from depth images using convolutional neural networks. Previous methods addressing this problem rely on manifold learning to learn low dimensional viewpoint descriptors and employ them in a nearest neighbor search on an estimated descriptor space. In comparison we create an efficient multi-task learning framework combining manifold descriptor learning and pose regression. By combining the strengths of manifold learning using triplet loss and pose regression, we could either estimate the pose directly reducing the complexity compared to NN search, or use the learned descriptor for the NN descriptor matching. By in depth experimental evaluation of the novel loss function we observed that the view descriptors learned by the network are much more discriminative resulting in almost 30% increase regarding relative pose accuracy compared to related works. On the other hand, regarding directly regressed poses we obtained important improvement compared to simple pose regression. By leveraging the advantages of both manifold learning and regression tasks, we are able to improve the current state-of-the-art for object recognition and pose retrieval.
15
Laurens van der Maaten, Geoffrey E. Hinton · Journal of Machine Learning Research · 2008 · 35.7K citations
Scene Coordinate Regression Forests for Camera Relocalization in RGB-D Images
Jamie Shotton, Ben Glocker, Christopher Zach et al. · 2013 · 951 citations
Uncertainty-Driven 6D Pose Estimation of Objects and Scenes from a Single RGB Image
Eric Brachmann, Frank Michel, Alexander Krull et al. · 2016 · 552 citations