A list of 3D point cloud completion resources. We try to keep it updated every week or two with the latest papers.

• [PoinTr] PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers, ICCV 2021, X. Yu et al. [PDF][Code]
• [Progressive Seed Generation] Progressive Seed Generation Auto-encoder for Unsupervised Point Cloud Learning, ICCV 2021, J. Yang et al. [PDF][[Code]]
• [RFNet] RFNet: Recurrent Forward Network for Dense Point Cloud Completion, ICCV 2021, T. Huang et al. [PDF][[Code]]
• [OcCo] Unsupervised Point Cloud Pre-training via Occlusion Completion, ICCV 2021, H. Wang et al. [PDF][Code]
• [SnowflakeNet] SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer, ICCV2021, P. Xiang et al. [PDF][Code]
• [VE-PCN] Voxel-based Network for Shape Completion by Leveraging Edge Generation, ICCV2021, X. Wang et al. [PDF][Code]
• [Calibrated Backprojection Network] Unsupervised Depth Completion with Calibrated Backprojection Layers, ICCV2021, A. Wang et al. [PDF][Code]
• [GarmentNets] GarmentNets: Category-Level Pose Estimation for Garments via Canonical Space Shape Completion, ICCV 2021, X. Yu et al. [PDF][[Code]]
• [ME-PCN] ME-PCN: Point Completion Conditioned on Mask Emptiness, ICCV 2021, B. Gong et al. [PDF][[Code]]
• [Point-Voxel Diffusion] 3D Shape Generation and Completion through Point-Voxel Diffusion, ICCV 2021, L. Zhou et al. [PDF][[Code]]
• [Bayesian Deep] Bayesian Deep Basis Fitting for Depth Completion with Uncertainty, ICCV2021, C. Qu et al. [PDF][[Code]]
• [Shape-Inversion] Unsupervised 3D Shape Completion through GAN Inversion, CVPR2021,J. Zhang et al. [PDF][Code]
• [Deco] Denoise and Contrast for Category Agnostic Shape Completion, CVPR2021, A. Alliegro et al. [PDF][Code]
• [Deco Supplementary] Denoise and Contrast for Category Agnostic Shape Completion Supplementary Material, CVPR2021, Julian Chibane et al. [PDF][[Code]]
• [Cycle4Completion] Cycle4Completion: Unpaired Point Cloud Completion using Cycle Transformation with Missing Region Coding, CVPR2021, X. Wen et al. [PDF][Code]
• [VRCNet] Variational Relational Point Completion Network, CVPR2021, L. Pan et al. [PDF][Code]
• [PMP-Net] PMP-Net: Point Cloud Completion by Learning Multi-step Point Moving Paths, CVPR2021, X. Wen et al. [PDF][Code]
• [Neural Tangent Kernel Perspective] Unsupervised Shape Completion via Deep Prior in the Neural Tangent Kernel Perspective, ACM TOG2021, L. Chu et al. [PDF]
• [ASFM-Net] ASFM-Net: Asymmetrical Siamese Feature Matching Network for Point Completion, ACM MM2021, Y. Xia et al. [PDF][Code]
• [ASHF-Net] ASHF-Net: Adaptive Sampling and Hierarchical Folding Network for Robust Point Cloud Completion, AAAI2021, D. Zong et al. [PDF][[Code]]
• [PC-RGNN] PC-RGNN: Point Cloud Completion and Graph Neural Network for 3D Object Detection, AAAI 2021, Y. Zhang et al. [PDF][[Code]]
• [Deep Learning] Point Cloud Completion by Deep Learning, publication 2021, R Kossat [PDF][[Code]]
• [Self-driving] An Efficient Deep Learning Approach Using Improved Generative Adversarial Networks for Incomplete Information Completion of Self-driving, arXiv, J. Tu et al. [PDF][[Code]]
• [TransSC] TransSC: Transformer-based Shape Completion for Grasp Evaluation, arXiv, W. Chen et al. [PDF][[Code]]
• [HyperPocket] HyperPocket: Generative Point Cloud Completion, arXiv, P. Spurek et al. [PDF][Code]
• [FinerPCN] FinerPCN: High fidelity point cloud completion network using pointwise convolution, Neucom2021, Y. Chang et al. [PDF][[Code]]
• [3D Grid Transformation] 3D Grid Transformation Network For Point Cloud Completion, ICIP2021, X. Deng et al. [PDF][[Code]]
• [STACKED AUTO-ENCODER] 3D POINT CLOUD COMPLETION USING STACKED AUTO-ENCODER FOR STRUCTURE PRESERVATION, ICIP2021, S. Kumari et al. [PDF][[Code]]
• [LPCC-Net] LPCC-Net: RGB Guided Local Point Cloud Completion for Outdoor 3D Object Detection, ICME2021, Y. Wei et al.[PDF][Code]
• [Novel Depth View Synthesis] Towards Efficient 3D Point Cloud Scene Completion via Novel Depth View Synthesis, ICPR2021, H. Wang et al.[PDF][[Code]]
• [Multiscale Feature Fusion] Point cloud completion using multiscale feature fusion and cross-regional attention, Image and Vision Computing2021, H. Wu et al. [PDF][[Code]]
• [Cross-Cascade Graph CNN] Towards point cloud completion: Point Rank Sampling and Cross-Cascade Graph CNN, Neurocomputing2021, L. Zhu et al.[PDF][[Code]]

• [PF-Net] PF-Net: Point Fractal Network for 3D Point Cloud Completion, CVPR2020, Z. Huang et al. [PDF][Code]
• [CRN] Cascaded Refinement Network for Point Cloud Completion with Self-supervision, CVPR2020, X. Wang et al. [PDF][Code]
• [SA-Net] Point Cloud Completion by Skip-attention Network with Hierarchical Folding, CVPR2020, X. Wen et al. [PDF][[Code]]
• [IF-Net] Implicit Functions in Feature Space for 3D Shape Reconstruction and Completion, CVPR2020, Julian Chibane et al. [PDF][Code]
• [O-CNN] Deep Octree-based CNNs with Output-Guided Skip Connections for 3D Shape and Scene Completion, CVPRW2020, PS. Wang et al. [PDF][Code]
• [SFA] Detail Preserved Point Cloud Completion via Separated Feature Aggregation, ECCV 2020, W. Zhang et al. [PDF][Code]
• [GRNet] GRNet: Gridding Residual Network for Dense Point Cloud Completion, ECCV2020, H. Xie et al. [PDF][Code]
• [SoftPoolNet] SoftPoolNet: Shape Descriptor for Point Cloud Completion and Classification, ECCV2020, Y. Wang et al. [PDF][Code]
• [PCL2PCL] Unpaired Point Cloud Completion on Real Scans using Adversarial Training, ICLR2020, X. Chen et al. [PDF][Code]
• [MSN] Morphing and Sampling Network for Dense Point Cloud Completion, AAAI2020, M. Liu et al. [PDF][Code]
• [Multi-View Consistent Inference] 3D Shape Completion with Multi-View Consistent Inference, AAAI2020, Tao Hu et al. [PDF][[Code]]
• [SAUM] SAUM: Symmetry-Aware Upsampling Module for Consistent Point Cloud Completion, ACCV2020, Hyeontae Son et al. [PDF][Code]
• [S3CNet] S3CNet: A Sparse Semantic Scene Completion Network for LiDAR Point Clouds, arXiv, R. Cheng et al. [PDF][[Code]]
• [PointSetVoting] Point Set Voting for Partial Point Cloud Analysis, arXiv, J. Zhang et al. [PDF][Code]
• [KAPLAN] KAPLAN: A 3D Point Descriptor for Shape Completion, 3DV2020, A. Richard et al.[PDF][[Code]]
• [Vaccine-style-net] Vaccine-style-net: Point Cloud Completion in Implicit Continuous Function Space,ACM MM2020, W. Yan et al. [PDF][[Code]]

• [3D Capsule] 3D Point Capsule Networks, CVPR2019, Y. Zhao et al. [PDF][Code]
• [TopNet] TopNet: Structural Point Cloud Decoder, CVPR2019, Lyne P. Tchapmi et al. [PDF][Code]
• [RL-GAN-Net] RL-GAN-Net: A Reinforcement Learning Agent Controlled GAN Network for Real-Time Point Cloud Shape Completion, CVPR2019, Muhammad Sarmad et al. [PDF][Code]
• [ShapeCompletion3DTracking] Leveraging Shape Completion for 3D Siamese Tracking, CVPR2019, S. Giancola et al. [PDF][Code]
• [Volume-guided Progressive View Inpainting] Deep Reinforcement Learning of Volume-guided Progressive View Inpainting for 3D Point Scene Completion from a Single Depth Image, CVPR2019, X. Han et al. [PDF][[Code]]
• [Multi-Angle Point Cloud-VAE] Multi-Angle Point Cloud-VAE: Unsupervised Feature Learning for 3D Point Clouds From Multiple Angles by Joint Self-Reconstruction and Half-to-Half Prediction, ICCV2019, Z. Han et al. [PDF][[Code]]
• [PU-GAN] PU-GAN: a Point Cloud Upsampling Adversarial Network, ICCV2019, R. Li et al. [PDF][Code]
• [Render4Completion] Render4Completion: Synthesizing Multi-View Depth Maps for 3D Shape Completion, ICCV2019, T. Hu et al. [PDF][[Code]]
• [Point-Cloud-Shape-Completion] High Fidelity Semantic Shape Completion for Point Clouds using Latent Optimization, WACV2019, S. Agrawal et al. [PDF][Code]

• [Daml-Shape-Completion] Learning 3D Shape Completion from Laser Scan Data with Weak Supervision, CVPR2018, David Stutz et al. [PDF][Code]
• [PU-Net] PU-Net: Point Cloud Upsampling Network, CVPR2018, L. Yu et al. [PDF][Code]
• [Single-View] Learning Shape Priors for Single-View 3D Completion and Reconstruction, ECCV2018, J. Wu et al. [PDF][[Code]]
• [PCN] PCN: Point completion network, 3DV2018, W. Yuan et al. [PDF][Code]

• [CNN Complete] Shape Completion using 3D-Encoder-Predictor CNNs and Shape Synthesis, CVPR2017, A. Dai et al. [PDF][Code]
• [PointSetGeneration] A Point Set Generation Network for 3D Object Reconstruction from a Single Image, CVPR2017, H. Fan et al. [PDF][Code]
• [High-Resolution] High-Resolution Shape Completion Using Deep Neural Networks for Global Structure and Local Geometry Inference, ICCV2017, X. Han et al. [PDF][[Code]]
• [Automatic Object Shape Completion] Automatic Object Shape Completion from 3D Point Clouds for Object Manipulation, ICCV2017, R. Figueiredo et al. [PDF][[Code]]
• [Quadrangulated Patches] Learning quadrangulated patches for 3D shape parameterization and completion, 3DV2017, K. Sarkar et al. [PDF][[Code]]
• [Shape-controllable] Shape-controllable geometry completion for point cloud models, The Visual Computer2017, L. Yang et al. [PDF][[Code]]

• [Heuristic 3D Object Shape Completion] Heuristic 3D Object Shape Completion based on Symmetry and Scene Context, IROS2016, D. Schiebener et al. [PDF][[Code]]
• [Single RGBD Image] Shape Completion from a Single RGBD Image,IEEE TVCG 2016, D. Li et al. [PDF][[Code]]
• [3D ShapeNets] 3D ShapeNets: A Deep Representation for Volumetric Shapes, CVPR2015, Z. Wu et al. [PDF][[Code]]
• [Data-driven structual priors] Data-driven structural priors for shape completion, ACM TOG2015,M. Sung et al. [PDF][[Code]]
• [3D-PatchMatch] 3D-PatchMatch: An optimization algorithm for point cloud completion, ICSDM2015, Z. Cai et al. [PDF][[Code]]
• [SCAPE] SCAPE: Shape Completion and Animation of People, ACM TOG2015, D. Anguelov et al. [PDF][[Code]]
• [Context-based Coherent Surface] Context-based Coherent Surface Completion, ACM TOG2014, G. Harary et al.[PDF][[Code]]
• [3D Aware Correction and Completion] 3D Aware Correction and Completion of Depth Maps in Piecewise Planar Scenes, ACCV 2014, A. K. Thabet et al. [PDF][[Code]]
• [Temporally Coherent Completion] Temporally Coherent Completion of Dynamic Shapes, ACM2012, H. Li et al.[PDF][[Code]]
• [Geometry Completion] Geometry completion and detail generation by texture synthesis, The Visual Computer2005, M.X. Nguyen et al.PDF[[Code]]
• [Non-parametric] Non-parametric 3D Surface Completion, 3DIM2005, Toby P. Breckon et al. [PDF][[Code]]
• [Example-Based] Example-Based 3D Scan Completion, SGP2005, M. Pauly et al. [[PDF]]

• PCN dataset
• CRN dataset
• ShapeNet Benchmark dataset
  • ShapeNet-55 Benchmark
  • ShapeNet-34 Benchmark
  • ShapeNet-Core dataset
  • Shapenet-Part dataset
• Completion3D benchmark dataset
• 3D-EPN dataset
• ModelNet dataset
  • ModelNet10 dataset
  • ModelNet40 dataset
• 3DMatch benchmark dataset
• S3DIS dataset
• PF-Net dataset
• PartNet dataset

• KITTI dataset
• ScanNet dataset
• Matterport3D dataset
• D-FAUST dataset

• Chamfer Distance (CD)
  • CD-T
  • CD-P
• Unidirectional Chamfer Distance (UCD)
• Unidirectional Hausdorff Distance (UHD)
• Total Mutual Difference (TMD)
• Fréchet Point Cloud Distance (FPD):FPD evaluates the distribution similarity by the 2-Wasserstein distance between the real and fake Gaussian measured in the feature spaces of the point sets.
• Earth Mover Distance (EMD)
• Accuracy: Accuracy measures the fraction of points in the output that are matched with the ground truth
• Completeness: Similar to accuracy, completeness reports the fraction of points in the ground truth that are within a distance threshold to any point in the output.
• F-score: F-score is calculated as the harmonic average of the accuracy and completeness.
• Fidelity. Fidelity measures how well the inputs are preserved in the outputs.
• Fidelity error：Fidelity error is the average distance from each point in the input to its nearest neighbour in the output.
• Consistency
• Plausibility. Plausibility is evaluated as the classification accuracy in percentage by a pre-trained PointNet model.
• Intersection over Union (IoU)
• Mean Intersection over Union(mIoU)
• JSD: The Jensen-Shannon Divergence between marginal distributions defined in the Euclidean 3D space.
• Coverage(COV):Coverage measures the fraction of point clouds in the reference set that is matched to at least one point cloud in the generated set. For each point cloud in the generated set, its nearest neighbor in the reference set is marked as a match.
• Minimum Matching Distance (MMD):MMD is proposed to complement coverage as a metric that measures quality.For each point cloud in the reference set, the distance to its nearest neighbor in the generated set is computed and averaged.
  • MMD-EMD
  • MMD-CD
• Point Moving Distance(PMD):minimize the sum of all displacement vector