Pith. sign in

REVIEW 3 major objections 4 minor 30 references

Skip priors and add graph-based anatomical information, for point-based Couinaud segmentation

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper claims that point-based Couinaud segmentation can match or beat vessel-prior methods without any explicit vessel prior, by learning anatomical structure through a 3D dynamic graph reasoning module.

desk verdict Useful architectural simplification, but the single unvalidated split and thin LiTS margin mean the 'exceed' claim is not yet supported. read the letter →

arxiv 2508.01785 v1 pith:T2MHG53D submitted 2025-08-03 cs.CV

Xiaotong Zhang , Alexander Broersen , Gonnie CM van Erp , Silvia L. Pintea , Jouke Dijkstra This is my paper · ORCID
classification cs.CV
keywords Couinaudsegmentation3Dgraphreasoningpoint-basedliverCTimplicitanatomicalinformationdynamicmessagepassingvessel-free
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

This paper sets out to show that point-based Couinaud segmentation of liver CT can be done accurately without the manual liver vessel prior that earlier point-based methods required. Couinaud segmentation divides the liver into eight functionally independent segments whose borders come from hepatic and portal vein planes, and it is used in planning liver resection and ablation. The proposed model learns that vessel structure implicitly: at each of four sampling levels it revoxelizes the point features and applies a dynamic graph reasoning module that computes affinities among neighboring voxels, so no explicit vessel labels are needed at inference. On the MSD and LiTS benchmarks it reports the highest average Dice among four compared point-based baselines, 80.41% versus 76.00% on MSD and 79.56% versus 78.62% on LiTS, with competitive average surface distance. The practical payoff is that surgical planning could skip the time-consuming vessel-annotation step without giving up segment accuracy.

What carries the argument

The load-bearing mechanism is the dynamic graph reasoning module $G_r(f(p_i))$ applied at each of four down-sampled point levels. The model first revoxelizes the point features into a $[M_i\times M_i\times M_i]$ grid and processes them with two 3D residual convolutions, producing a grid feature embedding $f(p_i)$ that bridges point space and grid space. In that grid, the module computes affinities over a fixed subset of $3^3=27$ voxels selected by learnable 3D offsets, with the update $$\hat{p}_i = \mathrm{softmax}\big(q_i\cdot \$\Delta$(k_i,\delta_i)+\mathrm{pos}_i\big)\cdot \$\Delta$(v_i,\delta_i),$$ where $q_i,k_i,v_i$ are self-attention projections and $\Delta(\cdot,\delta_i)$ is a deformable unfold that adapts keys and values to the offsets. The output is devoxelized back to point coordinates by interpolation. This mechanism carries the claim because it replaces the externally supplied vessel prior with affinities that the network itself learns from textural and positional patterns around the vessel planes.

What would settle it

Re-run the full method and its ablation on the same public data with the same inherited annotations and the same 10/3/7 split: if the 80.41 MSD and 79.56 LiTS average Dice are not reproduced, or if removing the graph reasoning module fails to lower Dice by roughly two points, the claim that implicit graph-based anatomy drives the gain is not supported.

Watch

Extended reading notes

Core claim

The central claim is that the anatomical information needed for Couinaud segmentation can be learned from the CT intensities themselves instead of being provided as a prior. The authors add a 3D dynamic graph reasoning module on top of voxelized point features; the module selects a small set of voxels with learnable 3D offsets and computes self-attention affinities among them, which together encode implicit anatomical relations between liver regions. With only eight segment labels as supervision, this module yields an average Dice of 80.41 on MSD and 79.56 on LiTS, higher than the vessel-prior point-based baseline at 76.00 and 78.62, and the best average surface distance on LiTS among the four baselines. The ablation study shows both the grid feature embeddings and the graph reasoning module are needed: removing the graph module drops average Dice by about two points on each dataset, and removing the grid embeddings drops it further. The paper's intended message is therefore that explicit vessel priors are not necessary for competitive point-based Couinaud segmentation, because a learned graph can supply the anatomy.

Load-bearing premise

The weakest load-bearing premise is that the Couinaud ground-truth labels inherited from prior work, together with the random 10/3/7 train/validation/test split, are accurate and representative enough that the reported Dice margins reflect the method rather than annotation bias or an unlucky split.

Editorial extensions

If this is right

  • Liver surgery planning can skip the manual vessel-prior step: the proposed model matches or exceeds a vessel-prior point-based baseline while trained only on segment labels.
  • The method keeps native CT resolution: points are placed at each voxel's physical coordinates, so volumes with axial spacing up to 7.5 mm (MSD) and down to 1.0 mm (LiTS) are handled without resampling to a fixed grid.
  • The graph module is doing real work: ablating $G_r(f(p))$ lowers average Dice by roughly two points on both datasets, so the reported gain depends on the learned affinities, not on the backbone alone.
  • Segment boundaries stay close to the expected vessel planes in the qualitative results, but the method is not robust to large tumors that erase landmarks such as the left hepatic vein or left portal vein, as the paper's failure analysis shows.

Reading between the lines

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

  • A natural next test is whether the learned 3D offsets align across patients with the actual hepatic and portal vein planes; if they do, the module could double as a weakly supervised vessel-anatomy detector, which the paper does not claim.
  • The same skip-prior recipe may transfer to other anatomy-defined partition tasks where boundaries are invisible planes, such as bronchopulmonary segments or brain vascular territories, though the paper only studies liver Couinaud segments.
  • Because the paper identifies large tumors as the main failure mode, an extension would be to condition the graph reasoning on tumor regions or suppress them before affinity computation; this is an inference, not part of the paper's experiments.
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 proposes a point-based network for Couinaud liver segmentation from CT that avoids explicit liver-vessel priors. The method builds on AGCNN and adds two components: grid feature embeddings that bridge point features into a voxel grid, and a 3D dynamic graph reasoning module that learns affinities among a small set of neighboring voxels. The authors evaluate on the MSD and LiTS datasets using Couinaud annotations inherited from prior works, compare against four point-based baselines, report Dice and average surface distance, and provide an ablation study. The central claim is that the method is competitive with, and in the conclusion is said to exceed, the four baselines without requiring explicit vessel structure priors.

Significance. If the empirical claims are supported, the contribution is useful: removing the need for manually defined vessel priors while retaining anatomical context through learned graph affinities is a sensible direction for point-based Couinaud segmentation. Strengths include the use of two public datasets, comparison with four relevant baselines, a clear ablation design, qualitative failure-case analysis, and released code. The ablation demonstrates that the grid-feature bridge is necessary for the graph module to work. However, the current evidence is not sufficient for the stated 'exceed' conclusion: the headline comparisons rest on a single undisclosed split without error bars or significance tests, and the annotations are inherited from prior work with no independent validation, including the same first author's earlier method that is also the strongest baseline.

major comments (3)
  1. [Section 4, Implementation details; Tables 1–2] The main comparison is a single point estimate per method on one train/validation/test split. The paper only reports the split ratio (10/3/7), not the exact subject assignment or random seed, and Tables 1 and 2 contain no error bars, confidence intervals, or significance tests. On LiTS, the margin over the vessel-prior baseline [28] is 0.94 Dice points; without per-subject variance and a paired statistical test, this margin is not evidence of superiority. Please report per-case mean and standard deviation over multiple random splits, or at least bootstrapping over test subjects, and disclose the exact split and seed. The standard deviations in Table 3 are not accompanied by a statement of what they are computed over and do not by themselves validate the headline comparison.
  2. [Section 4, Dataset description] The Couinaud ground-truth labels are inherited from Tian et al. [21] and Zhang et al. [28] rather than generated and validated for this study. No annotation protocol, inter-observer agreement, or independent anatomical verification is reported. Because [28] is also the strongest baseline and is a prior work by the same first author, there is a risk of circularity if those annotations were produced by a method trained on subjects that overlap the evaluation set, or if the labels encode a vessel model that [28] matches by construction. Please describe the provenance of the annotations, report inter-rater reliability on a subset, and verify that no subject overlap exists between the annotation-generation procedure and the training/validation/test split used here.
  3. [Section 5, Discussion and conclusion] The conclusion states that the method 'exceed[s] PointNet, PointNet++, AGCNN and Zhang et al.'s method', whereas the abstract claims only that the method is 'competitive'. The reported evidence supports at most the weaker claim. If the statistical and annotation-validation concerns above are not addressed, the conclusion should be softened to reflect the actual strength of the evidence.
minor comments (4)
  1. [Section 3.2] The notation '33(= 27) voxels' is ambiguous and should read '3^3 (= 27) voxels'.
  2. [Section 4.2 and Figure 4 caption] There are typos: 'precictions' should be 'predictions', and 'heaptic' should be 'hepatic'.
  3. [Section 4, Dataset description] The exact source, license, and availability of the Couinaud annotations from [21] and [28] should be stated, since the reader currently cannot obtain the same labels to reproduce the experiments.
  4. [Section 4.1] The inference-time and GFLOPs numbers would be more useful if the software versions of PyTorch, Torchmetrics, and MONAI were specified, as these can affect runtime measurements.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the graph-reasoning contribution is trained end-to-end and evaluated on held-out data; the only same-author dependency is the provenance of the Couinaud ground-truth annotations, which is a benchmark-validity concern rather than a circular derivation.

full rationale

The paper's central claim is architectural: a 3D dynamic graph reasoning module over point features can learn anatomical affinities implicitly and remove the need for explicit vessel priors. The derivation chain is explicit and self-contained: input points and CT intensities proceed through the AGCNN backbone, grid feature embeddings f(p), the graph reasoning module Gr(f(p)), and a two-layer MLP, with an eight-way cross-entropy loss. All parameters are learned end-to-end and evaluated on a held-out test portion of a 10/3/7 split, so no fitted constant or training label is renamed as a prediction. The four baselines (PointNet, PointNet++, AGCNN, and Zhang et al. [28]) are external methods run with their official implementations, and the reported Dice/ASD comparisons are direct output comparisons on the same liver masks. The one same-author dependency is the source of Couinaud annotations: MSD and LiTS do not contain native Couinaud labels, so the paper uses annotations from Tian et al. [21] and Zhang et al. [28], where [28] shares the first author with the present work. That is a legitimate provenance limitation for benchmark validity, and it is also one of the compared baselines, but it does not make the method's predictions equal to its inputs by construction. No self-definitional equation, fitted-input-as-prediction, imported uniqueness theorem, or ansatz-smuggled-via-citation pattern appears; the graph module's behavior is supervised by the segmentation labels, not by a prior vessel model. The paper is therefore not circular in the sense defined by the analysis, and the modest self-citation is not load-bearing for the architectural claim.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim does not depend on a single fitted physical constant. The free parameters are architecture hyperparameters, chosen by hand and standard for the inherited components (AGCNN and DGMN). The main hidden assumption is the validity of the externally sourced Couinaud labels. The method introduces no new physical entities, only a learned attention mechanism.

free parameters (4)
  • ball-query radius r = 1/(2*64) for MSD, 1/(2*32) for LiTS
    Hand-chosen to control the neighborhood size for neighbor sampling at the first scale; affects the receptive field but is not fitted to a target result.
  • first-scale grid size M_i = 64^3 (MSD), 32^3 (LiTS)
    Chosen by dataset resolution; affects the granularity of voxelized point features.
  • number of affinity voxels = 27 (3^3)
    From DGMN [27], the deformable attention samples 27 voxels; chosen for memory efficiency, not for a specific outcome.
  • training hyperparameters = 400 epochs, LR 0.01, momentum 0.98
    Standard training choices reported in Implementation details; no sensitivity analysis given.
assumptions (4)
  • domain assumption The liver can be divided into eight Couinaud segments determined by the hepatic and portal vein planes, and these segment boundaries are learnable from CT intensities and point coordinates without explicit vessel labels.
    This is the anatomical premise of the whole method, stated in the Introduction and Fig. 1.
  • domain assumption The ground-truth Couinaud annotations from Tian et al. [21] and Zhang et al. [28] are accurate and consistent for the MSD and LiTS scans.
    The evaluation relies entirely on these external labels; no independent validation is provided (Section 4).
  • ad hoc to paper The grid feature embeddings f(p) are a necessary bridge; graph reasoning alone on voxelized point features fails, as shown by ablation (c) in Table 3.
    The paper uses this design choice to justify the architecture, but it is an empirical finding, not a derived principle.
  • standard math Standard backpropagation and cross-entropy training on the provided labels will optimize the segmentation objective.
    Assumed throughout the Methods section; not stated explicitly.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Skip priors and add graph-based anatomical information, for point-based Couinaud segmentation." pith.science (2026). https://pith.science/paper/T2MHG53D

@misc{pith2026250801785,
  author       = {Pith},
  title        = {Pith review of: Skip priors and add graph-based anatomical information, for point-based Couinaud segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T2MHG53D}},
  note         = {Machine review of arXiv:2508.01785}
}
read the original abstract

The preoperative planning of liver surgery relies on Couinaud segmentation from computed tomography (CT) images, to reduce the risk of bleeding and guide the resection procedure. Using 3D point-based representations, rather than voxelizing the CT volume, has the benefit of preserving the physical resolution of the CT. However, point-based representations need prior knowledge of the liver vessel structure, which is time consuming to acquire. Here, we propose a point-based method for Couinaud segmentation, without explicitly providing the prior liver vessel structure. To allow the model to learn this anatomical liver vessel structure, we add a graph reasoning module on top of the point features. This adds implicit anatomical information to the model, by learning affinities across point neighborhoods. Our method is competitive on the MSD and LiTS public datasets in Dice coefficient and average surface distance scores compared to four pioneering point-based methods. Our code is available at https://github.com/ZhangXiaotong015/GrPn.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 24 canonical work pages

  1. [28]

    Computers in Biology and Medicine182, 109202 (2024) Title Suppressed Due to Excessive Length 11

    Zhang, X., Ali, S., Liu, T., Zhao, X., Cui, Z., Han, M., Ma, S., Zhu, J., Kang, Y., Wang, L., et al.: Robust and smooth couinaud segmentation via anatomical structure-guided point-voxel network. Computers in Biology and Medicine182, 109202 (2024) Title Suppressed Due to Excessive Length 11

    work page 2024

  2. [21]

    In: International workshop on machine learning in medical imaging

    Tian,J.,Liu,L.,Shi,Z.,Xu,F.:Automaticcouinaudsegmentationfromctvolumes on liver using glc-unet. In: International workshop on machine learning in medical imaging. pp. 274–282. Springer (2019)

    work page 2019

  3. [1]

    Journal of Digital Imaging33(2), 304–323 (2020)

    Alirr, O.I., Rahni, A.A.A.: Survey on liver tumour resection planning system: steps, techniques, and parameters. Journal of Digital Imaging33(2), 304–323 (2020)

    work page 2020

  4. [2]

    Nature communications13(1), 4128 (2022)

    Antonelli, M., Reinke, A., Bakas, S., Farahani, K., Kopp-Schneider, A., Landman, B.A., Litjens, G., Menze, B., Ronneberger, O., Summers, R.M., et al.: The medical segmentation decathlon. Nature communications13(1), 4128 (2022)

    2022

  5. [3]

    In: Medical Image Understanding and Analysis: 25th Annual Conference, MIUA 2021, Oxford, United Kingdom, July 12–14, 2021, Proceedings 25

    Arya, Z., Ridgway, G., Jandor, A., Aljabar, P.: Deep learning-based landmark lo- calisation in the liver for couinaud segmentation. In: Medical Image Understanding and Analysis: 25th Annual Conference, MIUA 2021, Oxford, United Kingdom, July 12–14, 2021, Proceedings 25. pp. 227–237. Springer (2021)

    work page 2021

  6. [4]

    Medical Image Analysis84, 102680 (2023)

    Bilic,P.,Christ,P.,Li,H.B.,Vorontsov,E.,Ben-Cohen,A.,Kaissis,G.,Szeskin,A., Jacobs, C., Mamani, G.E.H., Chartrand, G., et al.: The liver tumor segmentation benchmark (lits). Medical Image Analysis84, 102680 (2023)

    work page 2023

  7. [5]

    In: Medical Imaging 2006: Visualization, Image-Guided Procedures, and Display

    Boltcheva, D., Passat, N., Agnus, V., Jacob-Da Col, M.A., Ronse, C., Soler, L.: Automatic anatomical segmentation of the liver by separation planes. In: Medical Imaging 2006: Visualization, Image-Guided Procedures, and Display. vol. 6141, pp. 383–394. SPIE (2006)

    work page 2006

  8. [6]

    arXiv preprint arXiv:2211.02701 (2022)

    Cardoso, M.J., Li, W., Brown, R., Ma, N., Kerfoot, E., Wang, Y., Murrey, B., Myronenko, A., Zhao, C., Yang, D., et al.: Monai: An open-source framework for deep learning in healthcare. arXiv preprint arXiv:2211.02701 (2022)

    arXiv 2022

Show all 30 references
  1. [7]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Chen, Y., Rohrbach, M., Yan, Z., Shuicheng, Y., Feng, J., Kalantidis, Y.: Graph- based global reasoning networks. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 433–442 (2019)

    2019

  2. [8]

    Journal of Open Source Software7(70), 4101 (2022)

    Detlefsen, N.S., Borovec, J., Schock, J., Jha, A.H., Koker, T., Di Liello, L., Stancl, D., Quan, C., Grechkin, M., Falcon, W.: Torchmetrics-measuring reproducibility in pytorch. Journal of Open Source Software7(70), 4101 (2022)

    2022

  3. [9]

    KSII Transactions on Internet and Information Systems (TIIS)16(1), 16–37 (2022)

    Jia, X., Qian, C., Yang, Z., Xu, H., Han, X., Ren, H., Wu, X., Ma, B., Yang, D., Min, H.: Boundary-aware dual attention guided liver segment segmentation model. KSII Transactions on Internet and Information Systems (TIIS)16(1), 16–37 (2022)

    2022

  4. [10]

    Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., Guo, B.: Swin transformer:Hierarchical visiontransformerusingshiftedwindows.In:Proceedings of the IEEE/CVF international conference on computer vision. pp. 10012–10022 (2021)

    2021

  5. [11]

    Advances in neural information processing systems32 (2019)

    Liu, Z., Tang, H., Lin, Y., Han, S.: Point-voxel cnn for efficient 3d deep learning. Advances in neural information processing systems32 (2019)

    2019

  6. [12]

    Nature Reviews: Disease Primers7(1), 6 (2021) 10 Authors Suppressed Due to Excessive Length

    Llovet, J.M., Kelley, R.K., Villanueva, A., Singal, A.G., Pikarsky, E., Roayaie, S., Lencioni, R., Koike, K., Zucman-Rossi, J., Finn, R.S.: Hepatocellular carcinoma (primer). Nature Reviews: Disease Primers7(1), 6 (2021) 10 Authors Suppressed Due to Excessive Length

    2021

  7. [13]

    Computers in Biology and Medicine145, 105506 (2022)

    Luo, M., Jiang, H., Shi, T.: Multi-stage puncture path planning algorithm of abla- tion needles for percutaneous radiofrequency ablation of liver tumors. Computers in Biology and Medicine145, 105506 (2022)

    2022

  8. [14]

    In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision

    Ma, Y., Guo, Y., Liu, H., Lei, Y., Wen, G.: Global context reasoning for seman- tic segmentation of 3d point clouds. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. pp. 2931–2940 (2020)

    2020

  9. [15]

    Pat- tern Recognition97, 107000 (2020)

    Manessi, F., Rozza, A., Manzo, M.: Dynamic graph convolutional networks. Pat- tern Recognition97, 107000 (2020)

    2020

  10. [16]

    In: 2021 43rd Annual International Conference of the IEEE En- gineering in Medicine & Biology Society (EMBC)

    Pla-Alemany, S., Romero, J.A., Santabárbara, J.M., Aliaga, R., Maceira, A.M., Moratal, D.: Automatic multi-atlas liver segmentation and couinaud classification from ct volumes. In: 2021 43rd Annual International Conference of the IEEE En- gineering in Medicine & Biology Societ...

    2021

  11. [17]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Qi, C.R., Su, H., Mo, K., Guibas, L.J.: Pointnet: Deep learning on point sets for 3d classification and segmentation. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 652–660 (2017)

    2017

  12. [18]

    Advances in neural information processing systems 30 (2017)

    Qi, C.R., Yi, L., Su, H., Guibas, L.J.: Pointnet++: Deep hierarchical feature learn- ing on point sets in a metric space. Advances in neural information processing systems 30 (2017)

    2017

  13. [19]

    Medicina (Kaunas) 42(2), 98–106 (2006)

    Rutkauskas, S., Gedrimas, V., Pundzius, J., Barauskas, G., Basevicius, A.: Clinical and anatomical basis for the classification of the structural parts of liver. Medicina (Kaunas) 42(2), 98–106 (2006)

    2006

  14. [20]

    In: Walker, M., Ji, H., Stent, A

    Shaw, P., Uszkoreit, J., Vaswani, A.: Self-attention with relative position represen- tations. In: Walker, M., Ji, H., Stent, A. (eds.) Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologie...

    2018

  15. [22]

    Computers in Biology and Medicine144, 105363 (2022)

    Wang, M., Jin, R., Lu, J., Song, E., Ma, G.: Automatic ct liver couinaud segmen- tation based on key bifurcation detection with attentive residual hourglass-based cascaded network. Computers in Biology and Medicine144, 105363 (2022)

    2022

  16. [23]

    In: Pro- ceedings of the IEEE conference on computer vision and pattern recognition

    Wang, X., Girshick, R., Gupta, A., He, K.: Non-local neural networks. In: Pro- ceedings of the IEEE conference on computer vision and pattern recognition. pp. 7794–7803 (2018)

    2018

  17. [24]

    ACM Transactions on Graphics (tog)38(5), 1–12 (2019)

    Wang, Y., Sun, Y., Liu, Z., Sarma, S.E., Bronstein, M.M., Solomon, J.M.: Dynamic graph cnn for learning on point clouds. ACM Transactions on Graphics (tog)38(5), 1–12 (2019)

    2019

  18. [25]

    IEEE Transac- tions on Pattern Analysis and Machine Intelligence45(8), 9374–9392 (2023)

    Wei, M., Wei, Z., Zhou, H., Hu, F., Si, H., Chen, Z., Zhu, Z., Qiu, J., Yan, X., Guo, Y., et al.: Agconv: Adaptive graph convolution on 3d point clouds. IEEE Transac- tions on Pattern Analysis and Machine Intelligence45(8), 9374–9392 (2023)

    2023

  19. [26]

    Ye, X.: calflops: a flops and params calculate tool for neural networks in pytorch framework (2023), https://github.com/MrYxJ/calculate-flops.pytorch

    2023

  20. [27]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2022)

    Zhang, L., Chen, M., Arnab, A., Xue, X., Torr, P.H.: Dynamic graph message passing networks for visual recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence (2022)

    2022

  21. [29]

    Computers & Graphics 124, 104050 (2024)

    Zhou,W.,Wang,Q.,Jin,W.,Shi,X.,He,Y.:Graphtransformerfor3dpointclouds classification and semantic segmentation. Computers & Graphics 124, 104050 (2024)

    2024

  22. [30]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Zhu, X., Hu, H., Lin, S., Dai, J.: Deformable convnets v2: More deformable, bet- ter results. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 9308–9316 (2019)

    2019

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.