We refer to the proposed solution as SA3D, for Segment Anything in 3D. It is only required to provide a manual segmentation prompt ( e.g., rough points) for the target object in a single view, which is used to generate its 2D mask in this view with SAM.

Articulated objects, composed of multiple rigid parts connected by joints allowing rotational or translational motion, such as doors, cupboards and spectacles are ubiquitous in our daily lives.

The ability to sense and reconstruct 3D appearance and geometry is critical to applications in vision, graphics, and beyond.

Project page is available at https://ku-cvlab.github.io/DaRF/ .

A neural radiance field (NeRF) enables the synthesis of cutting-edge realistic novel view images of a 3D scene. It includes density and color fields to model the shape and radiance of a scene, respectively. Supervised by the photometric loss in an end-to-end training manner, NeRF inherently suffers from the shape-radiance ambiguity problem, i.e., it can perfectly fit training views but does not guarantee decoupling the two fields correctly.

However, point cloud registration methods struggle to achieve precise global pose estimation, whereas previous pose-free NeRFs overlook geometric consistency in global reconstruction.

The rendered results are with lowreconstruction fidelity for NeRF [1]and NeRF-W [7]only with6and12training views. For NeRF [1]with18training views, the result shows higher fidelity, but the undesired reflection is also finally rendered (labeled by green box).