Papers › Learning Multi-View Aggregation In the Wild for Large-Scale 3D Semantic Segmentation

Learning Multi-View Aggregation In the Wild for Large-Scale 3D Semantic Segmentation

15 Apr 2022CVPR 2022 1arXiv:2204.07548archive 2025-07-28

Damien Robert, Bruno Vallet, Loic Landrieu

Recent works on 3D semantic segmentation propose to exploit the synergy between images and point clouds by processing each modality with a dedicated network and projecting learned 2D features onto 3D points. Merging large-scale point clouds and images raises several challenges, such as constructing a mapping between points and pixels, and aggregating features between multiple views. Current methods require mesh reconstruction or specialized sensors to recover occlusions, and use heuristics to select and aggregate available images. In contrast, we propose an end-to-end trainable multi-view aggregation model leveraging the viewing conditions of 3D points to merge features from images taken at arbitrary positions. Our method can combine standard 2D and 3D networks and outperforms both 3D models operating on colorized point clouds and hybrid 2D/3D networks without requiring colorization, meshing, or true depth maps. We set a new state-of-the-art for large-scale indoor/outdoor semantic segmentation on S3DIS (74.7 mIoU 6-Fold) and on KITTI-360 (58.3 mIoU). Our full pipeline is accessible at https://github.com/drprojects/DeepViewAgg, and only requires raw 3D scans and a set of images and poses.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

drprojects/deepviewagg officialmentioned in papermentioned on GitHubpytorchNOASSERTION report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

3D Semantic SegmentationColorizationMultimodal Deep LearningSemantic Segmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
3D Semantic Segmentation KITTI-360 DeepViewAgg Model size 41.2M #1 of 8 Archive leaderboard report
3D Semantic Segmentation KITTI-360 DeepViewAgg mIoU Category 73.66 #1 of 8 Archive leaderboard report
3D Semantic Segmentation KITTI-360 DeepViewAgg miou 58.3 #1 of 8 Archive leaderboard report
3D Semantic Segmentation KITTI-360 DeepViewAgg miou Val 57.8 #1 of 8 Archive leaderboard report
3D Semantic Segmentation KITTI-360 MinkowskiNet Model size 37.9M #2 of 8 Archive leaderboard report
3D Semantic Segmentation KITTI-360 MinkowskiNet mIoU Category 74.08 #2 of 8 Archive leaderboard report
3D Semantic Segmentation KITTI-360 MinkowskiNet miou 53.92 #2 of 8 Archive leaderboard report
3D Semantic Segmentation KITTI-360 MinkowskiNet miou Val 54.2 #2 of 8 Archive leaderboard report
Semantic Segmentation S3DIS DeepViewAgg Mean IoU 74.7 #13 of 54 Archive leaderboard report
Semantic Segmentation S3DIS DeepViewAgg Number of params 41.2M #13 of 54 Archive leaderboard report
Semantic Segmentation S3DIS DeepViewAgg Params (M) 41.2 #13 of 54 Archive leaderboard report
Semantic Segmentation S3DIS DeepViewAgg mAcc 83.8 #13 of 54 Archive leaderboard report
Semantic Segmentation S3DIS DeepViewAgg oAcc 90.1 #13 of 54 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections