Pith. sign in

REVIEW 3 major objections 4 minor 49 references

Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Existing point cloud classifiers, trained on clean CAD models, fail to generalize to real-world scans, and a new dataset and background-aware networks begin to close that gap.

desk verdict ScanObjectNN is the real contribution here; the dataset is well built and the synthetic-to-real gap finding is significant, but the BGA method's state-of-the-art claim is not supported by an apples-to-apples comparison. read the letter →

arxiv 1908.04616 v2 pith:U5WEYVAQ submitted 2019-08-13 cs.CV

Mikaela Angelina Uy , Quang-Hieu Pham , Binh-Son Hua , Duc Thanh Nguyen , Sai-Kit Yeung This is my paper · ORCID
classification cs.CV
keywords pointcloudclassificationScanObjectNNreal-world3DscansbackgroundclutterpartialitydomaingeneralizationjointandsegmentationModelNet40benchmark
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that point cloud object classification is far from solved once objects come from real-world scans rather than clean CAD models. To support this, it introduces ScanObjectNN, a 15-class dataset of about 2,900 objects mined from scanned indoor scenes, with perturbation variants that add background clutter, translation, rotation, and scaling. Benchmarks show that all tested state-of-the-art methods trained on ModelNet40 score below 50% overall accuracy on ScanObjectNN, and even when trained on the real data, the best baseline reaches only 78.5% on the hardest variant. The paper identifies background clutter and object partiality as key open problems, and proposes background-aware networks (BGA-PN++ and BGA-DGCNN) that jointly predict class labels and foreground masks, reaching 80.2% on the hardest variant.

What carries the argument

The load-bearing mechanism is the background-aware classification network (BGA), built on PointNet++ and DGCNN backbones. It adds a segmentation branch that predicts a per-point foreground mask, fed by the feature vector from the classification branch before the final layer, and trains both branches jointly with total loss $L_{\text{total}} = L_{\text{class}} + \lambda L_{\text{seg}}$ where $\lambda = 0.5$. The segmentation supervision forces the network to ignore background points when forming class features, which is what improves accuracy on cluttered real-world objects. The dataset construction is the other key piece: objects are extracted from SceneNN and ScanNet instance masks, with a 'background' variant that keeps all points inside the ground-truth bounding box and perturbed variants (PB_T25, PB_T25_R, PB_T50_R, PB_T50_RS) that shift, rotate, and scale the box to simulate detector-level inputs.

What would settle it

Re-annotate a random sample of a few hundred ScanObjectNN objects with fresh human labels and measure agreement with the dataset's labels; substantial disagreement would mean the reported accuracies are built on unreliable ground truth.

Watch

Extended reading notes

Core claim

The central discovery is that the near-perfect accuracy reported on ModelNet40 does not transfer to real-world scanned objects: every benchmarked method, including PointNet, PointNet++, PointCNN, DGCNN, SpiderCNN, and 3DmFV, achieves under 50% overall accuracy on ScanObjectNN when trained on ModelNet40. Even with real-world training data, accuracy on the hardest variant (PB_T50_RS) saturates around 78.5% for existing methods, and the proposed background-aware models improve this by jointly learning to segment foreground from background while classifying. The paper also finds that training on real-world data generalizes to CAD models better than the reverse direction, and provides part-segmentation baselines on real chairs.

Load-bearing premise

The benchmark numbers depend on the accuracy of the ground-truth object labels and masks inherited from SceneNN and ScanNet and manually filtered; if those underlying annotations are noisy or biased, the measured accuracies and the claimed performance gap would be distorted.

Editorial extensions

If this is right

  • Real-world point cloud classification remains an open problem; synthetic-trained models should not be assumed deployable on depth-sensor data.
  • Adding a foreground-mask prediction branch to classification networks improves accuracy in cluttered scenes, on both the new real-world dataset and on ModelNet40 when trained on real data.
  • Training on real-world scans generalizes to CAD models better than the reverse, suggesting that real data provides a stronger training signal for shape understanding.
  • The perturbation variants of ScanObjectNN provide a controlled testbed for studying the effects of background clutter, translation, rotation, and scaling on classification.
  • Part-segmentation annotations on real-world chairs are feasible and can support future work on partial object classification.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The benchmark's 50% ceiling suggests that progress measured on ModelNet40 has been inflated by the dataset's cleanliness; future architecture comparisons may need a real-world component.
  • The BGA idea could be combined with point cloud completion or detection pipelines: a detector's rough bounding box naturally matches the PB_T50_RS setting, so mask prediction could improve downstream recognition.
  • A testable extension is to run the same joint classification-segmentation training with weakly supervised masks (e.g., from a classifier's attention) to see whether the improvement persists without pixel-level labels.
  • The paper's domain-gap asymmetry (real-to-synthetic transfers better than synthetic-to-real) suggests that synthetic data generation, not just architectures, is the bottleneck for real-world point cloud recognition.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper argues that point cloud object classification is still unsolved for real-world data, and introduces ScanObjectNN, a dataset of about 2,902 objects in 15 categories derived from SceneNN and ScanNet scene meshes, with several enrichment variants (OBJ_ONLY, OBJ_BG, PB_T25, PB_T25_R, PB_T50_R, PB_T50_RS) that add background and geometric perturbations. The authors benchmark six existing methods on this dataset and show that models trained on ModelNet40 generalize poorly to real scans (below 50% overall accuracy), while training and testing on ScanObjectNN still leaves a large gap to synthetic-data performance. They identify background, partiality, and synthetic-to-real generalization as open problems, and propose two background-aware networks, BGA-PN++ and BGA-DGCNN, which add a mask-prediction branch to PointNet++ and DGCNN and are trained with a combined classification and segmentation loss.

Significance. If the dataset is sound, it is a genuinely useful community resource: real-scanned point cloud object classification has been under-served compared with CAD benchmarks, and the paper provides a large, publicly released benchmark with multiple difficulty variants, baseline evaluations, and part annotations of chairs. The synthetic-to-real generalization result in Table 3 is striking and well supported by the reported numbers. The proposed method is simple and reproducible, and the paper explicitly releases code. However, the method evaluation in Table 9 does not currently isolate the claimed contribution because the BGA models receive mask supervision that the baselines do not; this confound must be addressed before the state-of-the-art claim can be accepted.

major comments (3)
  1. [Section 5.1, Eq. (L_total), and Table 9] The comparison in Table 9 does not control for supervision. BGA-PN++ and BGA-DGCNN are trained with L_total = L_class + 0.5*L_seg, receiving per-point foreground masks, while the vanilla PointNet++ and DGCNN baselines are trained with class labels only. The reported gains (80.2 vs 77.9 on PB_T50_RS; 52.6 vs 47.4 on ModelNet40) could therefore come entirely from the extra mask supervision or from the added segmentation head rather than from the proposed background-aware mechanism. Please add ablations that separate these factors: (i) train vanilla PointNet++ and DGCNN with the same auxiliary mask loss and segmentation head; (ii) train BGA-PN++ and BGA-DGCNN with the mask head present but without L_seg; and (iii) if the intended use is test-time masking, also report results when ground-truth or predicted masks are used to strip background before classification.
  2. [Section 4.2, Table 4] The main benchmark table reports only the 'hardest split' of the random train/test division, with full per-split results deferred to the supplementary material. Since the split is labeled as hardest after evaluation, the reader cannot determine from the main text whether the 78.5% ceiling and the relative rankings are representative or are an artifact of one favorable/unfavorable split. Please report the mean and standard deviation across all splits in the main text, or specify a fixed, pre-registered split-selection protocol; the per-split results should not be used only as a supplement.
  3. [Section 3.2.1, Data Collection] The validity of the entire benchmark rests on the quality of the object labels and instance masks inherited from SceneNN and ScanNet and then manually filtered by the authors, yet no label-quality validation is reported. If the underlying instance segmentation is noisy or biased toward particular object geometries, the measured accuracy gap between synthetic and real data would be distorted. Please provide a quantitative audit, such as re-annotation of a random subset by a second annotator with agreement statistics, or at least an analysis of how results change under stricter filtering criteria.
minor comments (4)
  1. [Section 3.2.1] There is an apparent copy-paste error in the text: 'semantic labelsOther works synthesize challenges on CAD data...' is inserted in the middle of the sentence describing the raw object attributes; the duplicated text should be removed or moved.
  2. [Section 5.2] The sentence 'Note that, in this evaluation all methods were trained on our i.e. PB T50 RS' is missing a word and should read 'trained on our dataset, i.e., PB T50 RS'.
  3. [Table 9] The header 'Ours ModelNet40' is ambiguous; the first column group should be labeled 'ScanObjectNN (PB_T50_RS)' or similar to clearly distinguish the two test sets.
  4. [Section 4.1] Please clarify in the main text whether the 'background removed' procedure for the ModelNet40-trained evaluation applies to all variants including OBJ_BG and the PB_* variants, and if so, how the background points are removed in each case.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's claims are empirical benchmark results and a multi-task network; the confounded baseline comparison is an experimental-design issue, not a circular reduction.

full rationale

The paper's derivation chain is empirical rather than formal: it constructs ScanObjectNN from existing scene datasets, benchmarks existing classifiers under multiple variants, and proposes BGA networks trained with an auxiliary mask loss. No step reduces to its own inputs by construction. The dataset labels and masks originate from SceneNN/ScanNet annotations, but these are external, publicly released annotations, and the benchmark's conclusions are tested on held-out scenes. BGA's use of mask supervision is additional training signal, not a renaming of the classification target; at test time masks are predicted, not provided, and the reported accuracies are on disjoint scenes. The comparison of BGA against vanilla baselines that do not receive mask supervision is an experimental confound, since the improvement could come from the extra supervision or the segmentation head, but that is a correctness/ablation concern rather than circularity. The self-citation to SceneNN, whose authors overlap with the present paper, is not load-bearing in a circularity sense: the dataset is externally falsifiable and was manually filtered. The paper also states its limitation that BGA 'requires object masks and background to be included in the data,' which is an honest scope condition, not a hidden circular input. Consequently, no circular step can be quoted from the paper, and the appropriate finding is no significant circularity.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claims rest on the quality of inherited scene labels and on reasonable design choices for perturbations and the loss weight. No new physical or mathematical entities are introduced.

free parameters (2)
  • lambda (loss weight for segmentation branch) = 0.5
    The total loss is L_total = L_class + lambda * L_seg with lambda set to 0.5 in Section 5.1. This is a manually chosen hyperparameter that affects the proposed method's accuracy, but not the benchmark dataset claim.
  • Perturbation magnitudes T25, T50, R, S = 25%, 50%, rotation, scaling
    Chosen by hand in Section 3.2.2 to define increasing difficulty levels. They are dataset design choices, not fitted to data, but they determine the performance gradient shown in Table 4.
assumptions (3)
  • domain assumption SceneNN and ScanNet instance labels are reliable enough for object classification after manual filtering.
    Section 3.2.1 explains that objects with inconsistent labels were fixed and ambiguous ones discarded, but the remaining labels are taken as ground truth.
  • domain assumption The hardest split is a fair representation of the dataset difficulty.
    Section 4.2 reports results on the 'hardest split' chosen among random splits; the paper assumes this split is representative and not a cherry-picked outlier, with full splits relegated to the supplementary material.
  • domain assumption Perturbations that retain at least 50% of original object points still represent the same object class.
    Section 3.2.2 discards perturbed objects with less than 50% of original points remaining, assuming this threshold preserves object identity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data." pith.science (2026). https://pith.science/paper/U5WEYVAQ

@misc{pith2026190804616,
  author       = {Pith},
  title        = {Pith review of: Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U5WEYVAQ}},
  note         = {Machine review of arXiv:1908.04616}
}
read the original abstract

Deep learning techniques for point cloud data have demonstrated great potentials in solving classical problems in 3D computer vision such as 3D object classification and segmentation. Several recent 3D object classification methods have reported state-of-the-art performance on CAD model datasets such as ModelNet40 with high accuracy (~92%). Despite such impressive results, in this paper, we argue that object classification is still a challenging task when objects are framed with real-world settings. To prove this, we introduce ScanObjectNN, a new real-world point cloud object dataset based on scanned indoor scene data. From our comprehensive benchmark, we show that our dataset poses great challenges to existing point cloud classification techniques as objects from real-world scans are often cluttered with background and/or are partial due to occlusions. We identify three key open problems for point cloud object classification, and propose new point cloud classification neural networks that achieve state-of-the-art performance on classifying objects with cluttered background. Our dataset and code are publicly available in our project page https://hkust-vgd.github.io/scanobjectnn/.

Figures

Figures reproduced from arXiv: 1908.04616 by the authors.

Figure 1
Figure 1. Sample objects from our dataset. well-segmented, and noise-free. In this experiment, we use the uniformly dense point cloud variant as preprocessed by Qi et al. [30]. Each point cloud is randomly sampled to 1024 points as input to the networks unless otherwise stated. The point clouds are centered at zero, and we use local co￾ordinates (x, y, z) normalized to [−1, 1] as point attributes. We follow the default train/… view at source ↗
Figure 2
Figure 2. Example objects from our dataset. the classification, e.g., a pen may be under-segmented with a table where it sits on and thus could be considered as a part of the table rather than a separate object. To study these fac￾tors, we introduce a variant of our dataset where objects are attached with background data (OBJ BG). We determine such background by using the ground truth axis-aligned ob￾ject bounding boxes. Spec… view at source ↗
Figure 3
Figure 3. An object in different perturbation variants. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: Part segmentation on the chair category. From top to [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 4
Figure 4. Figure 4: Confusion matrices of (a) 3DmFV [2], (b) PointNet [30], (c) SpiderCNN [44], (d) PointNet++ [32], (e) DGCNN [42] and (f) PointCNN [25] on our hardest PB T50 RS. This shows that there are no major ambiguity issues among object classes in our dataset. background elements …
Figure 6
Figure 6. Figure 6: Our proposed network [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Sample objects and their corresponding predicted masks [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 44 canonical work pages

  1. [1]

    Zamir, Helen Jiang, Ioannis Brilakis, Martin Fischer, and Silvio Savarese

    Iro Armeni, Ozan Sener, Amir R. Zamir, Helen Jiang, Ioannis Brilakis, Martin Fischer, and Silvio Savarese. 3d semantic parsing of large-scale indoor spaces. In CVPR, 2016. 2

    work page 2016

  2. [2]

    3dmfv: Three-dimensional point cloud classification in real- time using convolutional neural networks

    Yizhak Ben-Shabat, Michael Lindenbaum, and Anath Fischer. 3dmfv: Three-dimensional point cloud classification in real- time using convolutional neural networks. IEEE Robotics and Automation Letters, 2018. 2, 3, 5, 6, 7, 8

    work page 2018

  3. [3]

    Learning to drive from simulation without real world labels

    Alex Bewley, Jessica Rigley, Yuxuan Liu, Jeffrey Hawke, Richard Shen, Vinh-Dieu Lam, and Alex Kendall. Learning to drive from simulation without real world labels. In Interna- tional Conference on Robotics and Automation (ICRA), 2019. 2, 4

    work page 2019

  4. [4]

    Iqbal, and Eckehard Steinbach

    Dmytro Bobkov, Sili Chen, Ruiqing Jian, Muhammad Z. Iqbal, and Eckehard Steinbach. Noise-resistant deep learning for object classification in three-dimensional point clouds using a point pair descriptor. IEEE Robotics and Automation Letters, 2018. 2, 4

    work page 2018

  5. [5]

    Yale-cmu-berkeley dataset for robotic manipulation research

    Berk Calli, Arjun Singh, James Bruce, Aaron Walsman, Kurt Konolige, Siddhartha Srinivasa, Pieter Abbeel, and Aaron M Dollar. Yale-cmu-berkeley dataset for robotic manipulation research. International Journal of Robotics Research, 2017. 2

    work page 2017

  6. [6]

    Mitigation of effects of occlusion on object recognition with deep neural networks through low-level image completion

    Ben Chandler and Ennio Mingolla. Mitigation of effects of occlusion on object recognition with deep neural networks through low-level image completion. In Comp. Int. and Neu- rosc., 2016. 2, 4

    work page 2016

  7. [7]

    Chang, Thomas Funkhouser, Leonidas Guibas, Pat Hanrahan, Qixing Huang, Zimo Li, Silvio Savarese, Mano- lis Savva, Shuran Song, Hao Su, Jianxiong Xiao, Li Yi, and Fisher Yu

    Angel X. Chang, Thomas Funkhouser, Leonidas Guibas, Pat Hanrahan, Qixing Huang, Zimo Li, Silvio Savarese, Mano- lis Savva, Shuran Song, Hao Su, Jianxiong Xiao, Li Yi, and Fisher Yu. ShapeNet: An Information-Rich 3D Model Reposi- tory. Technical Report arXiv:1512.03012, Stanford University — Princeton University — Toyota Technological Institute at Chicago, 2015. 2

    arXiv 2015

  8. [8]

    A large dataset of object scans

    Sungjoon Choi, Qian-Yi Zhou, Stephen Miller, and Vladlen Koltun. A large dataset of object scans. arXiv:1602.02481,

    arXiv

Show all 49 references
  1. [9]

    Scannet: Richly- annotated 3d reconstructions of indoor scenes

    Angela Dai, Angel X Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, and Matthias Niessner. Scannet: Richly- annotated 3d reconstructions of indoor scenes. In CVPR,

  2. [10]

    Unsupervised feature learning for classi- fication of outdoor 3d scans

    Mark De Deuge, Alastair Quadros, Calvin Hung, and Bertrand Douillard. Unsupervised feature learning for classi- fication of outdoor 3d scans. In Australasian Conference on Robotics and Automation, 2013. 2

    2013

  3. [11]

    Dominguez, R

    M. Dominguez, R. Dhamdhere, A. Petkar, S. Jain, S. Sah, and R. Ptucha. General-purpose deep point cloud feature extractor. In WACV, 2018. 2

    2018

  4. [12]

    A study of the effect of noise and occlusion on the accuracy of convolutional neural networks applied to 3d object recogni- tion

    Alberto Garcia-Garcia, Jose Rodriguez, Sergio Orts, Sergiu Oprea, Francisco Gomez-Donoso, and Miguel Cazorla. A study of the effect of noise and occlusion on the accuracy of convolutional neural networks applied to 3d object recogni- tion. Computer Vision and Image Understandi...

    2017

  5. [13]

    Rich feature hierarchies for accurate object detection and semantic segmentation

    Ross Girshick, Jeff Donahue, Trevor Darrell, and Jitendra Malik. Rich feature hierarchies for accurate object detection and semantic segmentation. In CVPR, 2014. 4

    2014

  6. [14]

    Fabian Groh, Patrick Wieschollek, and Hendrik P. A. Lensch. Flex-convolution (million-scale point-cloud learning beyond grid-worlds). In ACCV, 2018. 3

    2018

  7. [15]

    Z. Han, H. Lu, Z. Liu, C. V ong, Y . Liua, M. Zwicker, J. Han, and C. L. P. Chen. 3d2seqviews: Aggregating sequential views for 3d global feature learning by cnn with hierarchical attention aggregation. IEEE Transactions on Image Process- ing, 2019. 2

    2019

  8. [16]

    View inter-prediction GAN: unsupervised repre- sentation learning for 3d shapes by learning global shape memories to support local view predictions

    Zhizhong Han, Mingyang Shang, Yu-Shen Liu, and Matthias Zwicker. View inter-prediction GAN: unsupervised repre- sentation learning for 3d shapes by learning global shape memories to support local view predictions. In AAAI, 2018. 2

    2018

  9. [17]

    Yˆ2seq2seq: Cross-modal representa- tion learning for 3d shape and text by joint reconstruction and prediction of view and word sequences

    Zhizhong Han, Mingyang Shang, Xiyang Wang, Yu-Shen Liu, and Matthias Zwicker. Yˆ2seq2seq: Cross-modal representa- tion learning for 3d shape and text by joint reconstruction and prediction of view and word sequences. In AAAI, 2019. 2

    2019

  10. [18]

    Hermosilla, T

    P. Hermosilla, T. Ritschel, P-P Vazquez, A. Vinacua, and T. Ropinski. Monte carlo convolution for learning on non- uniformly sampled point clouds. ACM Transactions on Graphics (Proceedings of SIGGRAPH Asia), 2018. 2

    2018

  11. [19]

    Scenenn: A scene meshes dataset with annotations

    Binh-Son Hua, Quang-Hieu Pham, Duc Thanh Nguyen, Minh- Khoi Tran, Lap-Fai Yu, and Sai-Kit Yeung. Scenenn: A scene meshes dataset with annotations. In International Conference on 3D Vision (3DV), 2016. http://www.scenenn.net. 1, 2, 3, 6, 7

    2016

  12. [20]

    Point- wise convolutional neural networks

    Binh-Son Hua, Minh-Khoi Tran, and Sai-Kit Yeung. Point- wise convolutional neural networks. In CVPR, 2018. 2

    2018

  13. [21]

    Phan, Bao Truong, Minh N

    Binh-Son Hua, Quang-Trung Truong, Minh-Khoi Tran, Quang-Hieu Pham, Asako Kanezaki, Tang Lee, HungYueh Chiang, Winston Hsu, Bo Li, Yijuan Lu, Henry Johan, Shoki Tashiro, Masaki Aono, Minh-Triet Tran, Viet-Khoi Pham, Hai-Dang Nguyen, Vinh-Tiep Nguyen, Quang-Thang Tran, Thuyen V ...

    2017

  14. [22]

    Rotationnet: Joint object categorization and pose estimation using multiviews from unsupervised viewpoints

    Asako Kanezaki, Yasuyuki Matsushita, and Yoshifumi Nishida. Rotationnet: Joint object categorization and pose estimation using multiviews from unsupervised viewpoints. In CVPR, 2018. 2

    2018

  15. [23]

    Lempitsky

    Roman Klokov and Victor S. Lempitsky. Escape from cells: Deep kd-networks for the recognition of 3d point cloud mod- els. 2017. 3

    2017

  16. [24]

    So-net: Self- organizing network for point cloud analysis

    Jiaxin Li, Ben M Chen, and Gim Hee Lee. So-net: Self- organizing network for point cloud analysis. In CVPR, 2018. 2, 3

    2018

  17. [25]

    Pointcnn: Convolution on x-transformed points

    Yangyan Li, Rui Bu, Mingchao Sun, and Baoquan Chen. Pointcnn: Convolution on x-transformed points. Advances in Neural Information Processing Systems, 2018. 2, 3, 5, 6, 7, 8

    2018

  18. [26]

    Point2sequence: Learning the shape representa- tion of 3d point clouds with an attention-based sequence to sequence network

    Xinhai Liu, Zhizhong Han, Yu-Shen Liu, and Matthias Zwicker. Point2sequence: Learning the shape representa- tion of 3d point clouds with an attention-based sequence to sequence network. arXiv:1811.02565, 2018. 2

    2018 arXiv

  19. [27]

    Maturana and S

    D. Maturana and S. Scherer. V oxNet: A 3D Convolutional Neural Network for Real-Time Object Recognition. In IROS,

  20. [28]

    A robust 3d-2d interactive tool for scene segmentation and annotation

    Duc Thanh Nguyen, Binh-Son Hua, Lap-Fai Yu, and Sai-Kit Yeung. A robust 3d-2d interactive tool for scene segmentation and annotation. IEEE Transactions on Visualization and Computer Graphics (TVCG), 2017. 2

    2017

  21. [29]

    Ninh, Tu-Khiem Le, Anh-Vu Dao, Vinh-Tiep Nguyen, Minh N

    Quang-Hieu Pham, Minh-Khoi Tran, Wenhui Li, Shu Xiang, Heyu Zhou, Weizhi Nie, Anan Liu, Yuting Su, Minh-Triet Tran, Ngoc-Minh Bui, Trong-Le Do, Tu V . Ninh, Tu-Khiem Le, Anh-Vu Dao, Vinh-Tiep Nguyen, Minh N. Do, Anh- Duc Duong, Binh-Son Hua, Lap-Fai Yu, Duc Thanh Nguyen, and S...

    2018

  22. [30]

    Pointnet: Deep learning on point sets for 3d classification and segmentation

    Charles R Qi, Hao Su, Kaichun Mo, and Leonidas J Guibas. Pointnet: Deep learning on point sets for 3d classification and segmentation. CVPR, 2017. 2, 3, 5, 6, 7, 8

    2017

  23. [31]

    Qi, Hao Su, Matthias Niessner, Angela Dai, Mengyuan Yan, and Leonidas J

    Charles R. Qi, Hao Su, Matthias Niessner, Angela Dai, Mengyuan Yan, and Leonidas J. Guibas. V olumetric and multi-view cnns for object classification on 3d data. In CVPR,

  24. [32]

    Point- net++: Deep hierarchical feature learning on point sets in a metric space

    Charles R Qi, Li Yi, Hao Su, and Leonidas J Guibas. Point- net++: Deep hierarchical feature learning on point sets in a metric space. Advances in Neural Information Processing Systems, 2017. 2, 3, 5, 6, 7, 8

    2017

  25. [33]

    Mining point cloud local structures by kernel correlation and graph pooling

    Yiru Shen, Chen Feng, Yaoqing Yang, and Dong Tian. Mining point cloud local structures by kernel correlation and graph pooling. In CVPR, 2018. 2, 3

    2018

  26. [34]

    Indoor segmentation and support inference from rgbd images

    Nathan Silberman, Derek Hoiem, Pushmeet Kohli, and Rob Fergus. Indoor segmentation and support inference from rgbd images. In ECCV, 2012. 2

    2012

  27. [35]

    Dynamic edge- conditioned filters in convolutional neural networks on graphs

    Martin Simonovsky and Nikos Komodakis. Dynamic edge- conditioned filters in convolutional neural networks on graphs. In CVPR, 2017. 3

    2017

  28. [36]

    Singh, J

    A. Singh, J. Sha, K. S. Narayan, T. Achim, and P. Abbeel. Bigbird: A large-scale 3d database of object instances. In In- ternational Conference on Robotics and Automation (ICRA),

  29. [37]

    Lichtenberg, and Jianxiong Xiao

    Shuran Song, Samuel P. Lichtenberg, and Jianxiong Xiao. Sun rgb-d: A rgb-d scene understanding benchmark suite. In CVPR, 2015. 2

    2015

  30. [38]

    Deep Sliding Shapes for amodal 3D object detection in RGB-D images

    Shuran Song and Jianxiong Xiao. Deep Sliding Shapes for amodal 3D object detection in RGB-D images. In CVPR,

  31. [39]

    Learned-Miller

    Hang Su, Subhransu Maji, Evangelos Kalogerakis, and Erik G. Learned-Miller. Multi-view convolutional neural networks for 3d shape recognition. In ICCV, 2015. 2

    2015

  32. [40]

    Semantic- paint: Interactive 3d labeling and learning at your fingertips

    Julien Valentin, Vibhav Vineet, Ming-Ming Cheng, David Kim, Jamie Shotton, Pushmeet Kohli, Matthias Nießner, An- tonio Criminisi, Shahram Izadi, and Philip Torr. Semantic- paint: Interactive 3d labeling and learning at your fingertips. ACM Transactions on Graphics, 2015. 6

    2015

  33. [41]

    Local spectral graph convolution for point set feature learning

    Chu Wang, Babak Samari, and Kaleem Siddiqi. Local spectral graph convolution for point set feature learning. ECCV, 2018. 2, 3

    2018

  34. [42]

    Sarma, Michael M

    Yue Wang, Yongbin Sun, Ziwei Liu, Sanjay E. Sarma, Michael M. Bronstein, and Justin M. Solomon. Dynamic graph cnn for learning on point clouds. arXiv preprint arXiv:1801.07829, 2018. 2, 3, 5, 6, 7, 8

    2018 arXiv

  35. [43]

    3d shapenets: A deep representation for volumetric shapes

    Zhirong Wu, Shuran Song, Aditya Khosla, Xiaoou Tang, and Jianxiong Xiao. 3d shapenets: A deep representation for volumetric shapes. In CVPR, 2015. 1, 2

    2015

  36. [44]

    Spidercnn: Deep learning on point sets with parameterized convolutional filters

    Yifan Xu, Tianqi Fan, Mingye Xu, Long Zeng, and Yu Qiao. Spidercnn: Deep learning on point sets with parameterized convolutional filters. In ECCV, 2018. 2, 3, 5, 6, 7, 8

    2018

  37. [45]

    Fold- ingnet: Point cloud auto-encoder via deep grid deformation

    Yaoqing Yang, Chen Feng, Yiru Shen, and Dong Tian. Fold- ingnet: Point cloud auto-encoder via deep grid deformation. In CVPR, 2018. 2

    2018

  38. [46]

    Spnet: Deep 3d object classification and retrieval using stereographic projection

    Mohsen Yavartanoo and Euyoung Kim. Spnet: Deep 3d object classification and retrieval using stereographic projection. In ACCV, 2018. 2

    2018

  39. [47]

    Pvnet: A joint convolutional network of point cloud and multi-view for 3d shape recognition

    Haoxuan You, Yifan Feng, Rongrong Ji, and Yue Gao. Pvnet: A joint convolutional network of point cloud and multi-view for 3d shape recognition. In Proceedings of the ACM Interna- tional Conference on Multimedia, 2018. 2

    2018

  40. [48]

    Multi-view har- monized bilinear network for 3d object recognition

    Tan Yu, Jingjing Meng, and Junsong Yuan. Multi-view har- monized bilinear network for 3d object recognition. In CVPR,

  41. [49]

    Deep sets

    Manzil Zaheer, Satwik Kottur, Siamak Ravanbakhsh, Barn- abas Poczos, Ruslan R Salakhutdinov, and Alexander J Smola. Deep sets. In Advances in Neural Information Processing Systems, 2017. 2, 3

    2017

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.