Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Panoptic Lifting for 3D Scene Understanding with Neural Fields

About

We propose Panoptic Lifting, a novel approach for learning panoptic 3D volumetric representations from images of in-the-wild scenes. Once trained, our model can render color images together with 3D-consistent panoptic segmentation from novel viewpoints. Unlike existing approaches which use 3D input directly or indirectly, our method requires only machine-generated 2D panoptic segmentation masks inferred from a pre-trained network. Our core contribution is a panoptic lifting scheme based on a neural field representation that generates a unified and multi-view consistent, 3D panoptic representation of the scene. To account for inconsistencies of 2D instance identifiers across views, we solve a linear assignment with a cost based on the model's current predictions and the machine-generated segmentation masks, thus enabling us to lift 2D instances to 3D in a consistent way. We further propose and ablate contributions that make our method more robust to noisy, machine-generated labels, including test-time augmentations for confidence estimates, segment consistency loss, bounded segmentation fields, and gradient stopping. Experimental results validate our approach on the challenging Hypersim, Replica, and ScanNet datasets, improving by 8.4, 13.8, and 10.6% in scene-level PQ over state of the art.

Yawar Siddiqui, Lorenzo Porzi, Samuel Rota Bul\'o, Norman M\"uller, Matthias Nie{\ss}ner, Angela Dai, Peter Kontschieder• 2022

Related benchmarks

TaskDatasetResultRank
Semantic segmentationReplica--
16
Panoptic SegmentationScanNet++
PQ (Panoptic Quality)71.14
14
Panoptic SegmentationScanNet V2
Panoptic Quality (PQ)57.86
14
3D Instance SegmentationScanNet
PQscene58.9
7
3D Instance SegmentationReplica3D
PQ Scene57.9
7
Semantic segmentationHyperSim v1 (test)
mIoU67.8
6
Semantic segmentationReplica v1 (test)
mIoU67.2
6
Semantic segmentationScanNet v1 (test)
mIoU65.2
6
Novel View SynthesisHyperSim v1 (test)
PSNR30.1
5
Novel View SynthesisScanNet v1 (test)
PSNR28.5
5
Showing 10 of 29 rows

Other info

Code

Follow for update