Skip to yearly menu bar Skip to main content


Poster

PruNeRF: Segment-Centric Dataset Pruning via 3D Spatial Consistency

Yeonsung Jung · Heecheol Yun · Joonhyung Park · Jin-Hwa Kim · Eunho Yang

Hall C 4-9 #802
[ ] [ Paper PDF ]
[ Poster
Thu 25 Jul 2:30 a.m. PDT — 4 a.m. PDT

Abstract:

Neural Radiance Fields (NeRF) have shown remarkable performance in learning 3D scenes. However, NeRF exhibits vulnerability when confronted with distractors in the training images -- unexpected objects are present only within specific views, such as moving entities like pedestrians or birds. Excluding distractors during dataset construction is a straightforward solution, but without prior knowledge of their types and quantities, it becomes prohibitively expensive. In this paper, we propose PruNeRF, a segment-centric dataset pruning framework via 3D spatial consistency, that effectively identifies and prunes the distractors. We first examine existing metrics for measuring pixel-wise distraction and introduce Influence Functions for more accurate measurements. Then, we assess 3D spatial consistency using a depth-based reprojection technique to obtain 3D-aware distraction. Furthermore, we incorporate segmentation for pixel-to-segment refinement, enabling more precise identification. Our experiments on benchmark datasets demonstrate that PruNeRF consistently outperforms state-of-the-art methods in robustness against distractors.

Chat is not available.