Variational Inference for Scalable 3D Object-centric Learning
This addresses the problem of scaling object-centric learning to larger 3D scenes for applications in computer vision and robotics, though it appears incremental by building on existing object-centric and NeRF approaches.
The paper tackles scalable unsupervised object-centric representation learning in 3D scenes by proposing a method that learns view-invariant object representations in localized coordinate systems, outperforming previous models in experiments on synthetic and real datasets.
We tackle the task of scalable unsupervised object-centric representation learning on 3D scenes. Existing approaches to object-centric representation learning show limitations in generalizing to larger scenes as their learning processes rely on a fixed global coordinate system. In contrast, we propose to learn view-invariant 3D object representations in localized object coordinate systems. To this end, we estimate the object pose and appearance representation separately and explicitly map object representations across views while maintaining object identities. We adopt an amortized variational inference pipeline that can process sequential input and scalably update object latent distributions online. To handle large-scale scenes with a varying number of objects, we further introduce a Cognitive Map that allows the registration and query of objects on a per-scene global map to achieve scalable representation learning. We explore the object-centric neural radiance field (NeRF) as our 3D scene representation, which is jointly modeled within our unsupervised object-centric learning framework. Experimental results on synthetic and real datasets show that our proposed method can infer and maintain object-centric representations of 3D scenes and outperforms previous models.