NeuralLabeling: A versatile toolset for labeling vision datasets using Neural Radiance Fields

TL;DR

NeuralLabeling is a versatile toolset that leverages Neural Radiance Fields to efficiently annotate complex 3D scenes from images, enabling improved dataset creation for robotic perception tasks.

Contribution

The paper introduces NeuralLabeling, a novel method using NeRFs for 3D scene annotation, including segmentation, bounding boxes, and poses, with demonstrated benefits in robotic grasping applications.

Findings

Higher reconstruction accuracy with supervised training on annotated depth maps.

Enhanced robotic grasping accuracy to 83.3% with depth completion.

Created the Dishwasher30k dataset with ground truth depth for transparent objects.

Abstract

We present NeuralLabeling, a labeling approach and toolset for annotating 3D scenes using either bounding boxes or meshes and generating segmentation masks, affordance maps, 2D bounding boxes, 3D bounding boxes, 6DOF object poses, depth maps, and object meshes. NeuralLabeling uses Neural Radiance Fields (NeRF) as a renderer, allowing labeling to be performed using 3D spatial tools while incorporating geometric clues such as occlusions, relying only on images captured from multiple viewpoints as input. To demonstrate the applicability of NeuralLabeling to a practical problem in robotics, we added ground truth depth maps to 30000 frames of transparent object RGB and noisy depth maps of glasses placed in a dishwasher captured using an RGBD sensor, yielding the Dishwasher30k dataset. We show that training a simple deep neural network with supervision using the annotated depth maps yields a higher reconstruction performance than training with the previously applied weakly supervised approach. We also show how instance segmentation and depth completion datasets generated using NeuralLabeling can be incorporated into a robot application for grasping transparent objects placed in a dishwasher with an accuracy of 83.3%, compared to 16.3% without depth completion.