mlsanityData to 14 Oct 2026

SCPNet: Semantic Scene Completion on Point Cloud

arXiv 2303.06884 ↗First posted 13 Mar 2023cs.CVcited by 80

On the map: Knowledge distillation → Detection and dense prediction distillation

Abstract (arXiv)

Training deep models for semantic scene completion (SSC) is challenging due to the sparse and incomplete input, a large quantity of objects of diverse scales as well as the inherent label noise for moving objects. To address the above-mentioned problems, we propose the following three solutions: 1) Redesigning the completion sub-network. We design a novel completion sub-network, which consists of several Multi-Path Blocks (MPBs) to aggregate multi-scale features and is free from the lossy downsampling operations. 2) Distilling rich knowledge from the multi-frame model. We design a novel knowledge distillation objective, dubbed Dense-to-Sparse Knowledge Distillation (DSKD). It transfers the dense, relation-based semantic knowledge from the multi-frame teacher to the single-frame student, significantly improving the representation learning of the single-frame model. 3) Completion label rectification. We propose a simple yet effective label rectification strategy, which uses off-the-shelf panoptic segmentation labels to remove the traces of dynamic objects in completion labels, greatly improving the performance of deep models especially for those moving objects. Extensive experiments are conducted in two public SSC benchmarks, i.e., SemanticKITTI and SemanticPOSS. Our SCPNet ranks 1st on SemanticKITTI semantic scene completion challenge and surpasses the competitive S3CNet by 7.2 mIoU. SCPNet also outperforms previous completion algorithms on the SemanticPOSS dataset. Besides, our method also achieves competitive results on SemanticKITTI semantic segmentation tasks, showing that knowledge learned in the scene completion is beneficial to the segmentation task.

Key facts. Cited by 80 papers in the map (data through 14 Oct 2026). Placed in Detection and dense prediction distillation, one of 12 directions in Knowledge distillation. Builds on 32 of its 42 references that are in the map. First posted on arXiv 13 Mar 2023.

Builds on

32 of 42 references.

Closest ancestors · 2

Same direction, newest first

  1. 2022Point-to-Voxel Knowledge Distillation for LiDAR Semantic Segmentation5 Jun
  2. 2022Towards Efficient 3D Object Detection with Knowledge Distillation30 May

Other references · 30

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  3. 2014FitNets: Hints for Thin Deep Netscited by 1,653
  4. 2016Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfercited by 1,077
  5. 2019Similarity-Preserving Knowledge Distillationcited by 516
  6. 2016Semantic Scene Completion from a Single Depth Imagecited by 472
  7. 2020Searching Efficient 3D Architectures with Sparse Point-Voxel Convolutioncited by 423
  8. 2020Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentationcited by 389
  9. 2020Sparse Single Sweep LiDAR Point Cloud Segmentation via Learning Contextual Shape Priors from Scene Completioncited by 223
  10. 2018Learning Deep Representations with Probabilistic Knowledge Transfercited by 213
  11. 2020LMSCNet: Lightweight Multiscale 3D Semantic Completioncited by 188
  12. 2021RPVNet: A Deep and Efficient Range-Point-Voxel Fusion Network for LiDAR Point Cloud Segmentationcited by 173
18 more
  1. 2021(AF)2-S3Net: Attentive Feature Fusion with Adaptive Feature Selection for Sparse Semantic Segmentation Networkcited by 164
  2. 2019Learning Lightweight Lane Detection CNNs by Self Attention Distillationcited by 162
  3. 20222DPASS: 2D Priors Assisted Semantic Segmentation on LiDAR Point Cloudscited by 158
  4. 2020SemanticPOSS: A Point Cloud Dataset with Large Quantity of Dynamic Instancescited by 123
  5. 2017ScanComplete: Large-Scale Scene Completion and Semantic Segmentation for 3D Scanscited by 103
  6. 2020S3CNet: A Sparse Semantic Scene Completion Network for LiDAR Point Cloudscited by 102
  7. 2020Semantic Scene Completion using Local Deep Implicit Functions on LiDAR Datacited by 74
  8. 20213D Semantic Scene Completion: a Surveycited by 74
  9. 2019Efficient Semantic Scene Completion Network with Spatial Group Convolutioncited by 66
  10. 2020Inter-Region Affinity Distillation for Road Marking Segmentationcited by 58
  11. 2021Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR-based Perceptioncited by 57
  12. 2018Two Stream 3D Semantic Scene Completioncited by 48
  13. 2022Comprehensive Review of Deep Learning-Based 3D Point Cloud Completion Processing and Analysiscited by 44
  14. 2022MotionSC: Data Set and Network for Real-Time Semantic Mapping in Dynamic Environmentscited by 40
  15. 2021Semantic Segmentation-assisted Scene Completion for LiDAR Point Cloudscited by 37
  16. 2018Learning to Steer by Mimicking Features from Heterogeneous Auxiliary Networkscited by 19
  17. 2021Categorical Relation-Preserving Contrastive Knowledge Distillation for Medical Image Classificationcited by 14
  18. 2022Mind the Gap in Distilling StyleGANscited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 11; 2024: 30; 2025: 25; 2026: 14.

80 papers.

Sort:

By area:

  1. 2024SparseOcc: Rethinking Sparse Latent Representation for Vision-Based Semantic Occupancy Predictioncited by 96
  2. 2024SemCity: Semantic Scene Generation with Triplane Diffusioncited by 53
  3. 2023PointOcc: Cylindrical Tri-Perspective View for Point-based 3D Semantic Occupancy Predictioncited by 52
  4. 2024Co-Occ: Coupling Explicit Feature Fusion with Volume Rendering Regularization for Multi-Modal 3D Semantic Occupancy Predictioncited by 48
  5. 2024MonoOcc: Digging into Monocular Semantic Occupancy Predictioncited by 44
  6. 2024Not All Voxels Are Equal: Hardness-Aware Semantic Scene Completion with Self-Distillationcited by 43
  7. 2023PaSCo: Urban 3D Panoptic Scene Completion with Uncertainty Awarenesscited by 42
  8. 2023Camera-based 3D Semantic Scene Completion with Sparse Guidance Networkcited by 40
  9. 2024RadarOcc: Robust 3D Occupancy Prediction with 4D Imaging Radarcited by 33
  10. 2025OmniNWM: Omniscient Driving Navigation World Modelscited by 33
  11. 2023DepthSSC: Monocular 3D Semantic Scene Completion via Depth-Spatial Alignment and Voxel Adaptationcited by 32
  12. 2024OccMamba: Semantic Occupancy Prediction with State Space Modelscited by 30
38 more
  1. 2023LiDAR-based 4D Occupancy Completion and Forecastingcited by 23
  2. 2024LiCROcc: Teach Radar for Accurate Semantic Occupancy Prediction using LiDAR and Cameracited by 19
  3. 2024DiffSSC: Semantic LiDAR Scan Completion using Denoising Diffusion Probabilistic Modelscited by 16
  4. 2023PointSSC: A Cooperative Vehicle-Infrastructure Point Cloud Benchmark for Semantic Scene Completioncited by 14
  5. 2025GaussianFormer3D: Multi-Modal Gaussian-based Semantic Occupancy Prediction with 3D Deformable Attentioncited by 14
  6. 2023TFNet: Exploiting Temporal Cues for Fast and Accurate LiDAR Semantic Segmentationcited by 13
  7. 2025VoxDet: Rethinking 3D Semantic Occupancy Prediction as Dense Object Detectioncited by 13
  8. 2024Offboard Occupancy Refinement with Hybrid Propagation for Autonomous Drivingcited by 12
  9. 2024TALoS: Enhancing Semantic Scene Completion via Test-time Adaptation on the Line of Sightcited by 12
  10. 2024WildOcc: A Benchmark for Off-Road 3D Semantic Occupancy Predictioncited by 10
  11. 2024Distilling Diffusion Models to Efficient 3D LiDAR Scene Completioncited by 10
  12. 2024OccScene: Semantic Occupancy-based Cross-task Mutual Learning for 3D Scene Generationcited by 10
  13. 2025LiDPM: Rethinking Point Diffusion for Lidar Scene Completioncited by 10
  14. 2024Label-efficient Semantic Scene Completion with Scribble Annotationscited by 9
  15. 2025Skip Mamba Diffusion for Monocular 3D Semantic Scene Completioncited by 9
  16. 2025PointSea: Point Cloud Completion via Self-structure Augmentationcited by 9
  17. 2025AGO: Adaptive Grounding for Open World 3D Occupancy Predictioncited by 9
  18. 2024Hierarchical Context Alignment with Disentangled Geometric and Temporal Modeling for Semantic Occupancy Predictioncited by 8
  19. 2025L2COcc: Lightweight Camera-Centric Semantic Scene Completion via Distillation of LiDAR Modelcited by 8
  20. 2025OneOcc: Semantic Occupancy Prediction for Legged Robots with a Single Panoramic Cameracited by 8
  21. 2024Real-time 3D semantic occupancy prediction for autonomous vehicles using memory-efficient sparse convolutioncited by 7
  22. 2025Pointmap Association and Piecewise-Plane Constraint for Consistent and Compact 3D Gaussian Segmentation Fieldcited by 7
  23. 2025Towards Learning to Complete Anything in Lidarcited by 7
  24. 2024Unsupervised 3D Point Cloud Completion via Multi-view Adversarial Learningcited by 6
  25. 2024FSF-Net: Enhance 4D Occupancy Forecasting with Coarse BEV Scene Flow for Autonomous Drivingcited by 6
  26. 2024ReliOcc: Towards Reliable Semantic Occupancy Prediction via Uncertainty Learningcited by 6
  27. 2025VLScene: Vision-Language Guidance Distillation for Camera-Based 3D Semantic Scene Completioncited by 6
  28. 2026OccAny: Generalized Unconstrained Urban 3D Occupancycited by 6
  29. 2023SPOT: Scalable 3D Pre-training via Occupancy Prediction for Learning Transferable 3D Representationscited by 5
  30. 2024UFO: Uncertainty-aware LiDAR-image Fusion for Off-road Semantic Terrain Map Estimationcited by 5
  31. 2024Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidancecited by 5
  32. 2025ProxyTransformation: Preshaping Point Cloud Manifold With Proxy Attention For 3D Visual Groundingcited by 5
  33. 2025Delta-Triplane Transformers as Occupancy World Modelscited by 5
  34. 2024Deep Learning for 3D Point Cloud Enhancement: A Surveycited by 4
  35. 2024Three Cars Approaching within 100m! Enhancing Distant Geometry by Tri-Axis Voxel Scanning for Camera-based Semantic Scene Completioncited by 4
  36. 2026Enhancing Indoor Occupancy Prediction via Sparse Query-Based Multi-Level Consistent Knowledge Distillationcited by 4
  37. 2023SCP: Scene Completion Pre-training for 3D Object Detectioncited by 3
  38. 2024ET-Former: Efficient Triplane Deformable Attention for 3D Semantic Scene Completion From Monocular Cameracited by 3

50 most cited of 80 All 80 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024Boosting Self-Supervision for Single-View Scene Completion via Knowledge Distillation
  2. 2025Feed-Forward SceneDINO for Unsupervised Semantic Scene Completion
  3. 2023S4C: Self-Supervised Semantic Scene Completion with Neural Fields
  4. 2025Imagine with the Teacher: Complete Shape in a Multi-View Distillation Way
  5. 2023Multi-to-Single Knowledge Distillation for Point Cloud Semantic Segmentation
  6. 2024Make a Strong Teacher with Label Assistance: A Novel Knowledge Distillation Approach for Semantic Segmentation
  7. 2026Heterogeneous and Adept Snapshot Distillation for 3D Semantic Segmentation
  8. 2023Fine-grained Text and Image Guided Point Cloud Completion with CLIP Model