Neural Radiance Field Codebooks
This work addresses the problem of enabling high-level scene understanding and efficient transfer to downstream tasks for AI and robotics applications, representing an incremental advancement with specific gains.
The paper tackles the challenge of learning compositional, object-centric representations for complex scenes by introducing Neural Radiance Field Codebooks (NRC), which uses a dictionary of object codes for novel view reconstruction. The result includes improvements such as a 3.1% higher success rate in object navigation, 29% relative improvement in unsupervised segmentation, and 5.5% better accuracy in depth ordering compared to prior methods.
Compositional representations of the world are a promising step towards enabling high-level scene understanding and efficient transfer to downstream tasks. Learning such representations for complex scenes and tasks remains an open challenge. Towards this goal, we introduce Neural Radiance Field Codebooks (NRC), a scalable method for learning object-centric representations through novel view reconstruction. NRC learns to reconstruct scenes from novel views using a dictionary of object codes which are decoded through a volumetric renderer. This enables the discovery of reoccurring visual and geometric patterns across scenes which are transferable to downstream tasks. We show that NRC representations transfer well to object navigation in THOR, outperforming 2D and 3D representation learning methods by 3.1% success rate. We demonstrate that our approach is able to perform unsupervised segmentation for more complex synthetic (THOR) and real scenes (NYU Depth) better than prior methods (29% relative improvement). Finally, we show that NRC improves on the task of depth ordering by 5.5% accuracy in THOR.