Papers › Point-LGMask: Local and Global Contexts Embedding for Point Cloud Pre-training with...

Point-LGMask: Local and Global Contexts Embedding for Point Cloud Pre-training with Multi-Ratio Masking

8 Jun 2023IEEE Transactions on Multimedia 2023 6archive 2025-07-28

Yuan Tang, Xianzhi Li, Jinfeng Xu, Qiao Yu, Long Hu, Yixue Hao, Min Chen

Self-supervised learning has achieved great success in both natural language processing and 2D vision, where masked modeling is a quite popular pre-training scheme. However, extending masking to 3D point cloud understanding that combines local and global features poses a new challenge. In our work, we present Point-LGMask, a novel method to embed both local and global contexts with multi-ratio masking, which is quite effective for self-supervised feature learning of point clouds but is unfortunately ignored by existing pre-training works. Specifically, to avoid fitting to a fixed masking ratio, we first propose multi-ratio masking, which prompts the encoder to fully explore representative features thanks to tasks of different difficulties. Next, to encourage the embedding of both local and global features, we formulate a compound loss, which consists of (i) a global representation contrastive loss to encourage the cluster assignments of the masked point clouds to be consistent to that of the completed input, and (ii) a local point cloud prediction loss to encourage accurate prediction of masked points. Equipped with our Point-LGMask, we show that our learned representations transfer well to various downstream tasks, including few-shot classification, shape classification, object part segmentation, as well as real-world scene-based 3D object detection and 3D semantic segmentation. Particularly, our model largely advances existing pre-training methods on the difficult few-shot classification task using the real-captured ScanObjectNN dataset by surpassing over 4% to the second-best method. Also, our Point-LGMask achieves 0.4% AP25 and 0.8% AP50 gains on 3D object detection task over the second-best method. 0.4% mAcc and 0.5% mIoU. Codes have been released at https://github.com/TangYuan96/Point-LGMask

PaperPDFCode

Code

TangYuan96/Point-LGMask mentioned in paperpytorch 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 Object Detection3D Point Cloud Classification3D Semantic SegmentationFew-Shot 3D Point Cloud ClassificationObject DetectionPoint Cloud Pre-trainingSelf-Supervised LearningSemantic Segmentationobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
3D Point Cloud Classification ScanObjectNN Point-LGMask OBJ-BG (OA) 89.8 #54 of 77 Archive leaderboard report
3D Point Cloud Classification ScanObjectNN Point-LGMask OBJ-ONLY (OA) 89.3 #54 of 77 Archive leaderboard report
3D Point Cloud Classification ScanObjectNN Point-LGMask Overall Accuracy 85.3 #54 of 77 Archive leaderboard report
Few-Shot 3D Point Cloud Classification ModelNet40 10-way (10-shot) Point-LGMask Overall Accuracy 92.6 #14 of 31 Archive leaderboard report
Few-Shot 3D Point Cloud Classification ModelNet40 10-way (10-shot) Point-LGMask Standard Deviation 4.3 #14 of 31 Archive leaderboard report
Few-Shot 3D Point Cloud Classification ModelNet40 10-way (20-shot) Point-LGMask Overall Accuracy 95.1 #14 of 31 Archive leaderboard report
Few-Shot 3D Point Cloud Classification ModelNet40 10-way (20-shot) Point-LGMask Standard Deviation 3.4 #14 of 31 Archive leaderboard report
Few-Shot 3D Point Cloud Classification ModelNet40 5-way (10-shot) Point-LGMask Overall Accuracy 97.4 #4 of 30 Archive leaderboard report
Few-Shot 3D Point Cloud Classification ModelNet40 5-way (10-shot) Point-LGMask Standard Deviation 2.0 #4 of 30 Archive leaderboard report
Few-Shot 3D Point Cloud Classification ModelNet40 5-way (20-shot) Point-LGMask Overall Accuracy 98.1 #12 of 30 Archive leaderboard report
Few-Shot 3D Point Cloud Classification ModelNet40 5-way (20-shot) Point-LGMask Standard Deviation 1.4 #12 of 30 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.

Methods

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

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