mlsanityData to 14 Oct 2026

MixSup: Mixed-grained Supervision for Label-efficient LiDAR-based 3D Object Detection

arXiv 2401.16305 ↗First posted 29 Jan 2024cs.CV cs.ROcited by 9

Not on the map. Why?

Abstract (arXiv)

Label-efficient LiDAR-based 3D object detection is currently dominated by weakly/semi-supervised methods. Instead of exclusively following one of them, we propose MixSup, a more practical paradigm simultaneously utilizing massive cheap coarse labels and a limited number of accurate labels for Mixed-grained Supervision. We start by observing that point clouds are usually textureless, making it hard to learn semantics. However, point clouds are geometrically rich and scale-invariant to the distances from sensors, making it relatively easy to learn the geometry of objects, such as poses and shapes. Thus, MixSup leverages massive coarse cluster-level labels to learn semantics and a few expensive box-level labels to learn accurate poses and shapes. We redesign the label assignment in mainstream detectors, which allows them seamlessly integrated into MixSup, enabling practicality and universality. We validate its effectiveness in nuScenes, Waymo Open Dataset, and KITTI, employing various detectors. MixSup achieves up to 97.31% of fully supervised performance, using cheap cluster annotations and only 10% box annotations. Furthermore, we propose PointSAM based on the Segment Anything Model for automated coarse labeling, further reducing the annotation burden. The code is available at https://github.com/BraveGroup/PointSAM-for-MixSup.

Key facts. Cited by 9 papers in the map (data through 14 Oct 2026). Builds on 48 of its 56 references that are in the map. First posted on arXiv 29 Jan 2024.

Builds on

48 of 56 references.

References · 48

  1. 2023Segment Anythingcited by 8,234
  2. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
  3. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
  4. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  5. 2019Scalability in Perception for Autonomous Driving: Waymo Open Datasetcited by 2,451
  6. 2020FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidencecited by 2,085
  7. 2017VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detectioncited by 1,323
  8. 2020Center-based 3D Object Detection and Trackingcited by 1,144
  9. 2018PointRCNN: 3D Object Proposal Generation and Detection from Point Cloudcited by 941
  10. 2019PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detectioncited by 891
  11. 2020PointContrast: Unsupervised Pre-training for 3D Point Cloud Understandingcited by 504
  12. 2019Hybrid Task Cascade for Instance Segmentationcited by 481
36 more
  1. 20203DSSD: Point-based 3D Single Stage Object Detectorcited by 440
  2. 2019From Points to Parts: 3D Object Detection from Point Cloud with Part-aware and Part-aggregation Networkcited by 368
  3. 2021PV-RCNN++: Point-Voxel Feature Set Abstraction With Local Vector Representation for 3D Object Detectioncited by 232
  4. 2023VoxelNeXt: Fully Sparse VoxelNet for 3D Object Detection and Trackingcited by 222
  5. 2021Embracing Single Stride 3D Object Detector with Sparse Transformercited by 193
  6. 2021Panoptic nuScenes: A Large-Scale Benchmark for LiDAR Panoptic Segmentation and Trackingcited by 166
  7. 2023SAM3D: Segment Anything in 3D Scenescited by 157
  8. 2021Offboard 3D Object Detection from Point Cloud Sequencescited by 145
  9. 2021LiDAR R-CNN: An Efficient and Universal 3D Object Detectorcited by 129
  10. 2023Segment Any Point Cloud Sequences by Distilling Vision Foundation Modelscited by 125
  11. 2023DSVT: Dynamic Sparse Voxel Transformer with Rotated Setscited by 122
  12. 2021Panoptic-PolarNet: Proposal-free LiDAR Point Cloud Panoptic Segmentationcited by 107
  13. 2019SESS: Self-Ensembling Semi-Supervised 3D Object Detectioncited by 89
  14. 20203DIoUMatch: Leveraging IoU Prediction for Semi-Supervised 3D Object Detectioncited by 89
  15. 2022Fully Sparse 3D Object Detectioncited by 84
  16. 2023UniTR: A Unified and Efficient Multi-Modal Transformer for Bird's-Eye-View Representationcited by 74
  17. 2023FlatFormer: Flattened Window Attention for Efficient Point Cloud Transformercited by 73
  18. 2022Scribble-Supervised LiDAR Semantic Segmentationcited by 69
  19. 2020Weakly Supervised 3D Object Detection from Lidar Point Cloudcited by 67
  20. 2023Towards Label-free Scene Understanding by Vision Foundation Modelscited by 54
  21. 2022Semi-supervised 3D Object Detection with Proficient Teacherscited by 45
  22. 2022ProposalContrast: Unsupervised Pre-training for LiDAR-based 3D Object Detectioncited by 43
  23. 2021Auto4D: Learning to Label 4D Objects from Sequential Point Cloudscited by 42
  24. 2020Weakly Supervised 3D Object Detection from Point Cloudscited by 41
  25. 2022LESS: Label-Efficient Semantic Segmentation for LiDAR Point Cloudscited by 41
  26. 2023SAM3D: Zero-Shot 3D Object Detection via Segment Anything Modelcited by 40
  27. 2021GP-S3Net: Graph-based Panoptic Sparse Semantic Segmentation Networkcited by 37
  28. 2023Super Sparse 3D Object Detectioncited by 32
  29. 2023DetZero: Rethinking Offboard 3D Object Detection with Long-term Sequential Point Cloudscited by 31
  30. 2023Once Detected, Never Lost: Surpassing Human Performance in Offline LiDAR based 3D Object Detectioncited by 28
  31. 2023MV-JAR: Masked Voxel Jigsaw and Reconstruction for LiDAR-Based Self-Supervised Pre-Trainingcited by 26
  32. 2023Hierarchical Supervision and Shuffle Data Augmentation for 3D Semi-Supervised Object Detectioncited by 24
  33. 2022Sparse Cross-scale Attention Network for Efficient LiDAR Panoptic Segmentationcited by 22
  34. 2021SMAC-Seg: LiDAR Panoptic Segmentation via Sparse Multi-directional Attention Clusteringcited by 11
  35. 2023ClickSeg: 3D Instance Segmentation with Click-Level Weak Annotationscited by 7
  36. 2023Center Focusing Network for Real-Time LiDAR Panoptic Segmentationcited by 6

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 3; 2025: 3; 2026: 3.

9 papers.

  1. 2026NeoVerse: Enhancing 4D World Model with in-the-wild Monocular Videoscited by 46
  2. 2024Delving into Multi-modal Multi-task Foundation Models for Road Scene Understanding: From Learning Paradigm Perspectivescited by 23
  3. 2025Learning to Detect Objects from Multi-Agent LiDAR Scans without Manual Labelscited by 7
  4. 2025SP3D: Boosting Sparsely-Supervised 3D Object Detection via Accurate Cross-Modal Semantic Promptscited by 6
  5. 2024Learning 3D Perception from Others' Predictionscited by 4
  6. 2025LiDAR Remote Sensing Meets Weak Supervision: Concepts, Methods, and Perspectivescited by 4
  7. 2026Few-Shot Incremental 3D Object Detection in Dynamic Indoor Environmentscited by 2
  8. 2024SC3D: Label-Efficient Outdoor 3D Object Detection via Single Click Annotationcited by 0
  9. 2026When the City Teaches the Car: Label-Free 3D Perception from Infrastructurecited by 0