Papers › Decoupled Local Aggregation for Point Cloud Learning

Decoupled Local Aggregation for Point Cloud Learning

31 Aug 2023arXiv:2308.16532archive 2025-07-28

Binjie Chen, Yunzhou Xia, Yu Zang, Cheng Wang, Jonathan Li

The unstructured nature of point clouds demands that local aggregation be adaptive to different local structures. Previous methods meet this by explicitly embedding spatial relations into each aggregation process. Although this coupled approach has been shown effective in generating clear semantics, aggregation can be greatly slowed down due to repeated relation learning and redundant computation to mix directional and point features. In this work, we propose to decouple the explicit modelling of spatial relations from local aggregation. We theoretically prove that basic neighbor pooling operations can too function without loss of clarity in feature fusion, so long as essential spatial information has been encoded in point features. As an instantiation of decoupled local aggregation, we present DeLA, a lightweight point network, where in each learning stage relative spatial encodings are first formed, and only pointwise convolutions plus edge max-pooling are used for local aggregation then. Further, a regularization term is employed to reduce potential ambiguity through the prediction of relative coordinates. Conceptually simple though, experimental results on five classic benchmarks demonstrate that DeLA achieves state-of-the-art performance with reduced or comparable latency. Specifically, DeLA achieves over 90\% overall accuracy on ScanObjectNN and 74\% mIoU on S3DIS Area 5. Our code is available at https://github.com/Matrix-ASC/DeLA .

PaperPDFCode

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

Code

matrix-asc/dela officialmentioned in papermentioned on GitHubpytorch 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 Point Cloud ClassificationSemantic SegmentationSupervised Only 3D Point Cloud Classification

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
3D Point Cloud Classification ModelNet40 DeLA FLOPs 1.44G #26 of 111 Archive leaderboard report
3D Point Cloud Classification ModelNet40 DeLA Mean Accuracy 92.2 #26 of 111 Archive leaderboard report
3D Point Cloud Classification ModelNet40 DeLA Number of params 5.3M #26 of 111 Archive leaderboard report
3D Point Cloud Classification ModelNet40 DeLA Overall Accuracy 94.0 #26 of 111 Archive leaderboard report
3D Point Cloud Classification ScanObjectNN DeLA FLOPs 1.5G #16 of 77 Archive leaderboard report
3D Point Cloud Classification ScanObjectNN DeLA Mean Accuracy 89.3 #16 of 77 Archive leaderboard report
3D Point Cloud Classification ScanObjectNN DeLA Number of params 5.3M #16 of 77 Archive leaderboard report
3D Point Cloud Classification ScanObjectNN DeLA Overall Accuracy 90.4 #16 of 77 Archive leaderboard report
Semantic Segmentation S3DIS Area5 DeLA Number of params 7.0M #6 of 61 Archive leaderboard report
Semantic Segmentation S3DIS Area5 DeLA mAcc 80.0 #6 of 61 Archive leaderboard report
Semantic Segmentation S3DIS Area5 DeLA mIoU 74.1 #6 of 61 Archive leaderboard report
Semantic Segmentation S3DIS Area5 DeLA oAcc 92.2 #6 of 61 Archive leaderboard report
Semantic Segmentation ScanNet DeLA val mIoU 75.9 #16 of 45 Archive leaderboard report
Supervised Only 3D Point Cloud Classification ScanObjectNN DeLA GFLOPs 1.5 #3 of 12 Archive leaderboard report
Supervised Only 3D Point Cloud Classification ScanObjectNN DeLA Number of params (M) 5.3 #3 of 12 Archive leaderboard report
Supervised Only 3D Point Cloud Classification ScanObjectNN DeLA Overall Accuracy (PB_T50_RS) 90.4 #3 of 12 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