NeuralViewSynthesisandMatching forSemi-SupervisedFew-ShotLearningof3DPose
–Neural Information Processing Systems
Ourmodel is trained in an EM-type manner alternating between increasing the 3D pose invariance ofthefeature extractor andannotating unlabelled data through neural viewsynthesis andmatching.
Neural Information Processing Systems
Feb-8-2026, 06:54:52 GMT