pith. sign in

arxiv: 2606.23653 · v1 · pith:LJ7XWDVMnew · submitted 2026-06-22 · 💻 cs.CV

Lightweight Neural Framework for Robust 3D Volume and Surface Estimation from Multi-View Images

Diego E. Farchione , Ramzi Idoughi , Peter Wonka This is my paper

Pith reviewed 2026-06-26 09:12 UTC · model grok-4.3

classification 💻 cs.CV
keywords volume estimationsurface areamulti-view imagesneural regressiongraph decoder3D reconstructionfeed-forward networkuncertainty estimation
0
0 comments X

The pith

A feed-forward network fuses point clouds and 2D features in a graph decoder to regress scale-normalized volume and surface area directly from multi-view images.

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

The paper introduces a lightweight neural framework that predicts volume and surface area values plus uncertainties straight from a set of images. It first builds a 3D point cloud, then merges that data with aligned 2D image features inside a graph decoder for immediate regression. The design removes the need for slow iterative optimization steps. Performance gains appear most clearly when only a few views are supplied. Results are shown on coral monitoring, food analysis, and body measurements.

Core claim

Fusing 3D point cloud reconstructions with view-aligned 2D features through a graph-based decoder enables accurate direct regression of scale-normalized volume and surface area, together with uncertainty estimates, from multi-view images without iterative optimization.

What carries the argument

Graph-based decoder that fuses 3D point cloud reconstructions with view-aligned 2D features to regress the target quantities.

If this is right

  • Inference runs in a single forward pass, enabling real-time use where optimization loops are too slow.
  • Accuracy holds with fewer input images than methods that require dense views or refinement steps.
  • Uncertainty outputs accompany each prediction, supporting downstream decision making in applied settings.
  • The same architecture applies across domains such as ecology, nutrition, and anthropometry without domain-specific tuning.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same fusion pattern could be tested for predicting additional shape descriptors such as principal axes or bounding-box dimensions.
  • Replacing the point-cloud front end with a learned implicit representation might further reduce dependence on explicit 3D reconstruction quality.
  • Deployment on edge devices could be measured directly by reporting frames-per-second and memory footprint on mobile hardware.

Load-bearing premise

That fusing point clouds and 2D features inside the graph decoder supplies enough information for direct regression to match or exceed optimization-based accuracy without extra constraints.

What would settle it

Run the model on a held-out test set of objects imaged from only two or three noisy views and compare absolute error in volume and surface area against an iterative optimization baseline; larger errors would falsify the performance claim.

Figures

Figures reproduced from arXiv: 2606.23653 by Diego E. Farchione, Peter Wonka, Ramzi Idoughi.

Figure 1
Figure 1. Figure 1: Teaser: Our end-to-end pipeline for robust volume and surface area estimation from multi-view images is demonstrated on corals from the CoralVOS dataset [58]: Using only five top-view RGB images from monocular video, our model can predict normalized volume and surface with their corresponding uncertainty, showcasing its potential for efficient, real-world applications. arXiv:2606.23653v1 [cs.CV] 22 Jun 202… view at source ↗
Figure 2
Figure 2. Figure 2: From masked multi-view images from meshes (or original images), a 3D re￾constructor like MapAnything produces a fused point cloud with per-point confidence, while a frozen encoder extracts view-aligned 2D features. A lightweight decoder sum￾marizes the point cloud, followed by max pooling layer. In the 2D branch, per-view features pass through a small FC block and are mean-max pooled across views. The fusi… view at source ↗
Figure 3
Figure 3. Figure 3: Representative samples across datasets. Top: food items from the MetaFood dataset. Middle: human subjects from THuman2.1. Bottom: synthetic coral specimens generated with Infinigen. These examples highlight the geometric and visual diversity of the domains on which our framework performs unified volume and surface estimation. To ensure reliable ground-truth geometry, all 3D meshes across our datasets are f… view at source ↗
Figure 4
Figure 4. Figure 4: Absolute Error vs. Uncertainty. Correlation between absolute error and predicted aleatoric/epistemic uncertainty for Evidential (left) and Eviden￾tial+Deterministic (right) models. Higher uncertainty aligns with larger errors [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Risk–Coverage (Total Uncertainty). MAE decreases as high-uncertainty samples are rejected. Both models outperform random rejection; Eviden￾tial+Deterministic shows improved calibration at low coverage [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Examples of 3D assets from the Objaverse dataset used in our pipeline, illus￾trating the diversity of shapes and textures. evenly spaced over 360^\circ azimuth, elevations 10^\circ \text {--}80^\circ , intensity 0.18 ; shading is PBR-style with ambient 0.2 , diffuse 0.9 , specular 0.4 , and full opacity. This setup ensures varied lighting and viewpoints while maintaining object focus. Note that this proces… view at source ↗
Figure 7
Figure 7. Figure 7: Images of collected real corals (first row), and snapshots of synthetic corals (second row). main paper and depicted in [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Example of segmentation inconsistencies in the MetaFood dataset. While some masks correctly isolate the full object (right), others segment only the food portion inside a container (left). This ambiguity arises because, in certain classes, the ground￾truth volume refers to both the container and its contents, while the mask includes only the food. Manual quality assurance (QA) was therefore required to ens… view at source ↗
Figure 9
Figure 9. Figure 9: Left: ground-truth image. Right: Agisoft reconstruction using 30 images, which fails to recover the correct geometry and appearance in most cases. well even from a single image, and more robust as the number of views increases; while Agisoft and Trellis either cannot be evaluated in some settings or remain significantly less accurate (see [PITH_FULL_IMAGE:figures/full_fig_p027_9.png] view at source ↗
read the original abstract

Accurate volume and surface area estimation is critical for diverse applications, from marine ecology to medical diagnostics. However, existing methods often suffer from high computational costs and poor performance with sparse and noisy data. We propose a fully feed-forward framework that regresses scale-normalized volume and surface area and their associated uncertainties directly from multi-view images. By fusing 3D point cloud reconstructions with view-aligned 2D features through a graph-based decoder, our model bypasses iterative optimization, ensuring exceptional scalability and rapid inference. Experimental results demonstrate that our approach outperforms state-of-the-art methods, particularly when operating with a low number of input images. Validated across coral monitoring, dietary analysis, and anthropometry, our proposed framework provides a robust, adaptable solution for quantitative shape analysis. This architecture provides a high-speed, scalable alternative for precise geometric estimation from visual data, maintaining high performance even in resource-constrained or sparse-view scenarios.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The manuscript proposes a fully feed-forward neural framework that directly regresses scale-normalized 3D volume and surface area (plus uncertainties) from multi-view images. It fuses initial 3D point-cloud reconstructions with view-aligned 2D features inside a graph-based decoder, thereby avoiding iterative optimization. The central claim is that this architecture outperforms state-of-the-art methods, especially under sparse-view conditions, and is validated on coral monitoring, dietary analysis, and anthropometry tasks.

Significance. If the performance and robustness claims are substantiated by quantitative experiments, the work would supply a lightweight, single-pass alternative for geometric estimation in resource-limited or sparse-view regimes, with direct applicability to ecology and medical imaging. The explicit uncertainty output and avoidance of iterative refinement constitute potential practical strengths.

major comments (2)
  1. [Abstract] Abstract: the claim that the approach 'outperforms state-of-the-art methods, particularly when operating with a low number of input images' is unsupported by any quantitative results, baselines, error metrics, error bars, or experimental protocol within the provided manuscript text.
  2. [Method] Method description (graph-based decoder): no derivation, error bound, or consistency constraint is supplied showing that fusion of noisy sparse point clouds with 2D features suffices for accurate scale-normalized regression without iterative refinement or geometric priors; this assumption is load-bearing for the central claim of robustness under sparse noisy inputs.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the major comments point-by-point below, with clarifications on experimental support and methodological details, and indicate planned revisions.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that the approach 'outperforms state-of-the-art methods, particularly when operating with a low number of input images' is unsupported by any quantitative results, baselines, error metrics, error bars, or experimental protocol within the provided manuscript text.

    Authors: The abstract summarizes findings from the experiments section, which includes quantitative comparisons against baselines with error metrics under varying view counts. However, to ensure the claim is fully supported and self-contained within the text, we will revise the abstract to reference key metrics, error bars, and the sparse-view protocol explicitly. revision: yes

  2. Referee: [Method] Method description (graph-based decoder): no derivation, error bound, or consistency constraint is supplied showing that fusion of noisy sparse point clouds with 2D features suffices for accurate scale-normalized regression without iterative refinement or geometric priors; this assumption is load-bearing for the central claim of robustness under sparse noisy inputs.

    Authors: The decoder learns the fusion end-to-end via supervision on ground-truth volumes and surfaces; consistency is implicitly enforced by the regression losses rather than explicit geometric priors. We will expand the method section with additional design rationale and discussion of the empirical robustness under sparse inputs, while noting the absence of formal bounds as a limitation of the data-driven approach. revision: partial

Circularity Check

0 steps flagged

No circularity in derivation; empirical performance claims independent of self-referential fits

full rationale

The paper describes a feed-forward neural architecture that fuses point clouds with 2D features via a graph decoder to regress scale-normalized volume/surface and uncertainties. No equations or steps in the provided text reduce a claimed prediction to a fitted parameter by construction, nor invoke self-citations as load-bearing uniqueness theorems. Performance claims are presented as experimental results against SOTA, not as mathematical identities derived from the model's own inputs. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no explicit free parameters, axioms, or invented entities are identifiable. The approach implicitly relies on standard neural-network training assumptions and the sufficiency of multi-view image data for the regression task, none of which are detailed.

pith-pipeline@v0.9.1-grok · 5694 in / 1132 out tokens · 30884 ms · 2026-06-26T09:12:58.819449+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

58 extracted references · 9 canonical work pages · 3 internal anchors

  1. [1]

    arXiv preprint arXiv:2407.01717 (2024)

    AlMughrabi, A., Haroon, U., Marques, R., Radeva, P.: VolETA: One-and few-shot food volume estimation. arXiv preprint arXiv:2407.01717 (2024)

    work page arXiv 2024

  2. [2]

    In: CVPRW

    AlMughrabi, A., Haroon, U., Marques, R., Radeva, P.: VolTex: Food volume es- timation using text-guided segmentation and neural surface reconstruction. In: CVPRW. pp. 450–457 (2025)

    2025

  3. [3]

    Advances in neural information processing systems33, 14927–14937 (2020)

    Amini, A., Schwarting, W., Soleimany, A., Rus, D.: Deep evidential regression. Advances in neural information processing systems33, 14927–14937 (2020)

    2020

  4. [4]

    In: ICCV

    Barron, J.T., Mildenhall, B., Tancik, M., Hedman, P., Martin-Brualla, R., Srini- vasan, P.P.: Mip-NeRF: A multiscale representation for anti-aliasing neural radi- ance fields. In: ICCV. pp. 5835–5844 (2021)

    2021

  5. [5]

    arXiv preprint arXiv:2409.01966 (2024)

    Chen, Y., He, J., Vinod, G., Raghavan, S., Czarnecki, C., Ma, J., Mahmud, T.I., Coburn, B., Mao, D., Nair, S., et al.: MetaFood3D: 3D food dataset with nutrition values. arXiv preprint arXiv:2409.01966 (2024)

    work page arXiv 2024

  6. [6]

    IEEE TMM19(5), 1090–1099 (2017)

    Dehais, J., Anthimopoulos, M., Shevchik, S., Mougiakakou, S.: Two-view 3d recon- struction for food volume estimation. IEEE TMM19(5), 1090–1099 (2017)

    2017

  7. [7]

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

    Deitke, M., Schwenk, D., Salvador, J., Weihs, L., Michel, O., VanderBilt, E., Schmidt, L., Ehsani, K., Kembhavi, A., Farhadi, A.: Objaverse: A universe of annotated 3d objects. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 13142–13153 (2023)

    2023

  8. [8]

    In: ICIP

    Fang, S., Zhu, F., Jiang, C., Zhang, S., Boushey, C.J., Delp, E.J.: A comparison of food portion size estimation using geometric models and depth images. In: ICIP. pp. 26–30 (2016)

    2016

  9. [9]

    Sensors23(6), 3298 (2023)

    Filko, D., Nyarko, E.K.: 2D/3D wound segmentation and measurement based on a robot-driven reconstruction system. Sensors23(6), 3298 (2023)

    2023

  10. [10]

    IEEE TPAMI32(8), 1362–1376 (2010)

    Furukawa, Y., Ponce, J.: Accurate, dense, and robust multiview stereopsis. IEEE TPAMI32(8), 1362–1376 (2010)

    2010

  11. [11]

    Gonzalez, B., Garcia, G., Gecele, O., Ramirez, J., Velastin, S.A., Farias, G.: Prelim- inary results on food weight estimation with RGB-D images. In: Int. Conf. Pattern Recog. Syst. pp. 1–7 (2024)

    2024

  12. [12]

    In: NIPS

    Hamilton, W., Ying, Z., Leskovec, J.: Inductive representation learning on large graphs. In: NIPS. p. 1024–1034 (2017)

    2017

  13. [13]

    The International Archives of the Pho- togrammetry,RemoteSensingandSpatialInformationSciences48,121–128(2024)

    Helmholz, P., Bassett, T., Boyle, L., Browne, N., Parnum, I., Moustaka, M., Evans, R.: Evaluating linear coral growth estimation using photogrammetry and alter- native point cloud comparison methods. The International Archives of the Pho- togrammetry,RemoteSensingandSpatialInformationSciences48,121–128(2024)

    2024

  14. [14]

    IEEE Trans

    Hu, P., Dai, X., Zhao, R., Wang, H., Ma, Y., Munteanu, A.: Point2partvolume: Human body volume estimation from a single depth image. IEEE Trans. Instrum. Meas.72, 1–12 (2023)

    2023

  15. [15]

    In: ACM SIGGRAPH 2024 conference papers

    Huang, B., Yu, Z., Chen, A., Geiger, A., Gao, S.: 2D gaussian splatting for geo- metrically accurate radiance fields. In: ACM SIGGRAPH 2024 conference papers. pp. 1–11 (2024)

    2024

  16. [16]

    arXiv preprint arXiv:2005.11621 (2020)

    Huang, J., Zhou, Y., Guibas, L.: ManifoldPlus: A robust and scalable wa- tertight manifold surface generation method for triangle soups. arXiv preprint arXiv:2005.11621 (2020)

    work page arXiv 2005

  17. [17]

    In: ECCV

    Jiang, Q., Li, F., Zeng, Z., Ren, T., Liu, S., Zhang, L.: T-Rex2: Towards generic object detection via text-visual prompt synergy. In: ECCV. pp. 38–57 (2024)

    2024

  18. [18]

    Measurement198, 111294 (2022) 16

    Kadam, P., Pandya, S., Phansalkar, S., Sarangdhar, M., Petkar, N., Kotecha, K., Garg, D.: FVEstimator: A novel food volume estimator wellness model for calorie measurement and healthy living. Measurement198, 111294 (2022) 16

    2022

  19. [19]

    In: International Con- ference on 3D Vision (3DV) (2026)

    Keetha, N., Müller, N., Schönberger, J., Porzi, L., Zhang, Y., Fischer, T., Knapitsch, A., Zauss, D., Weber, E., Antunes, N., Luiten, J., Lopez-Antequera, M., Bulò, S.R., Richardt, C., Ramanan, D., Scherer, S., Kontschieder, P.: MapA- nything: Universal feed-forward metric 3D reconstruction. In: International Con- ference on 3D Vision (3DV) (2026)

    2026

  20. [20]

    ACM TOG42(4), 139:1–139:14 (2023)

    Kerbl, B., Kopanas, G., Leimkühler, T., Drettakis, G.: 3D gaussian splatting for real-time radiance field rendering. ACM TOG42(4), 139:1–139:14 (2023)

    2023

  21. [21]

    In: ICLR (2017)

    Kipf, T.N., Welling, M.: Semi-supervised classification with graph convolutional networks. In: ICLR (2017)

    2017

  22. [22]

    In: ICCV

    Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., Lo, W.Y., Dollár, P., Girshick, R.: Segment anything. In: ICCV. pp. 3992–4003 (2023)

    2023

  23. [23]

    Frontiers in Marine Science8, 623645 (2021)

    Koch, H.R., Wallace, B., DeMerlis, A., Clark, A.S., Nowicki, R.J.: 3D scanning as a tool to measure growth rates of live coral microfragments used for coral reef restoration. Frontiers in Marine Science8, 623645 (2021)

    2021

  24. [24]

    Methods in Ecol- ogy and Evolution11(6), 714–726 (2020)

    Lange, I.D., Perry, C.T.: A quick, easy and non-invasive method to quantify coral growth rates using photogrammetry and 3D model comparisons. Methods in Ecol- ogy and Evolution11(6), 714–726 (2020)

    2020

  25. [25]

    arXiv preprint arXiv:2107.02259 (2021)

    Leinen, F., Cozzolino, V., Schön, T.: VolNet: Estimating human body part volumes from a single RGB image. arXiv preprint arXiv:2107.02259 (2021)

    work page arXiv 2021

  26. [26]

    BMC bioinformatics20(1), 724 (2019)

    Liu, C., Fan, X., Guo, Z., Mo, Z., Chang, E.I.C., Xu, Y.: Wound area measurement with 3d transformation and smartphone images. BMC bioinformatics20(1), 724 (2019)

    2019

  27. [27]

    In: ECCV

    Liu, S., Zeng, Z., Ren, T., Li, F., Zhang, H., Yang, J., Jiang, Q., Li, C., Yang, J., Su, H., Zhu, J., Zhang, L.: Grounding DINO:Marrying DINO with grounded pre-training for open-set object detection. In: ECCV. vol. 15105, pp. 38–55 (2024)

    2024

  28. [28]

    In: ACM MM

    Liu, X., Li, J., Zhang, X., Sun, J., Fan, X., Tian, Y.: Learning visible surface area estimation for irregular objects. In: ACM MM. pp. 2333–2343 (2022)

    2022

  29. [29]

    Loper, M., Mahmood, N., Romero, J., Pons-Moll, G., Black, M.J.: SMPL: A skinned multi-person linear model, pp. 851–866. ACM (2023)

    2023

  30. [30]

    IJCV60(2), 91–110 (2004)

    Lowe, D.G.: Distinctive image features from scale-invariant keypoints. IJCV60(2), 91–110 (2004)

    2004

  31. [31]

    In: CVPRW

    Lunscher, N., Zelek, J.: Deep learning whole body point cloud scans from a single depth map. In: CVPRW. pp. 1208–12087 (2018)

    2018

  32. [32]

    In: ICLR (2022)

    Ma, X., Qin, C., You, H., Ran, H., Fu, Y.: Rethinking network design and local geometry in point cloud: A simple residual MLP framework. In: ICLR (2022)

    2022

  33. [33]

    Mildenhall, B., Srinivasan, P.P., Tancik, M., Barron, J.T., Ramamoorthi, R., Ng, R.:Nerf:Representingscenesasneuralradiancefieldsforviewsynthesis.In:ECCV. vol. 12346, pp. 405–421 (2020)

    2020

  34. [34]

    Mirzaalian Dastjerdi, H., Töpfer, D., Rupitsch, S.J., Maier, A.: Measuring surface area of skin lesions with 2D and 3D algorithms. Int. J. Biomed. Imaging2019, 4035148:1–4035148:9 (2019)

    2019

  35. [35]

    ACM TOG41(4), 1–15 (2022)

    Müller, T., Evans, A., Schied, C., Keller, A.: Instant-NGP: Instant neural graphics primitives with a multiresolution hash encoding. ACM TOG41(4), 1–15 (2022)

    2022

  36. [36]

    In: ACM International Conference on Multimedia in Asia

    Naritomi, S., Yanai, K.: Hungry networks: 3D mesh reconstruction of a dish and a plate from a single dish image for estimating food volume. In: ACM International Conference on Multimedia in Asia. pp. 22:1–22:7 (2021)

    2021

  37. [37]

    European journal of clinical nutrition70(11), 1265–1270 (2016) 17

    Ng, B.K., Hinton, B.J., Fan, B., Kanaya, A.M., Shepherd, J.A.: Clinical anthropo- metrics and body composition from 3d whole-body surface scans. European journal of clinical nutrition70(11), 1265–1270 (2016) 17

    2016

  38. [38]

    In: CVPR

    Qi, C.R., Su, H., Mo, K., Guibas, L.J.: PointNet: Deep learning on point sets for 3d classification and segmentation. In: CVPR. pp. 77–85 (2017)

    2017

  39. [39]

    Qi, C.R., Yi, L., Su, H., Guibas, L.J.: Pointnet++: Deep hierarchical feature learn- ing on point sets in a metric space. In: Adv. Neural Inform. Process. Syst. pp. 5099–5108 (2017)

    2017

  40. [40]

    In: CVPR

    Raistrick, A., Lipson, L., Ma, Z., Mei, L., Wang, M., Zuo, Y., Kayan, K., Wen, H., Han, B., Wang, Y., Newell, A., Law, H., Goyal, A., Yang, K., Deng, J.: Infinite photorealistic worlds using procedural generation. In: CVPR. pp. 12630–12641 (2023)

    2023

  41. [41]

    In: ICLR (2025)

    Ravi, N., Gabeur, V., Hu, Y., Hu, R., Ryali, C., Ma, T., Khedr, H., Rädle, R., Rolland, C., Gustafson, L., Mintun, E., Pan, J., Alwala, K.V., Carion, N., Wu, C., Girshick, R.B., Dollár, P., Feichtenhofer, C.: SAM 2: Segment anything in images and videos. In: ICLR (2025)

    2025

  42. [42]

    org/abs/2405.10300

    Ren, T., Jiang, Q., Liu, S., Zeng, Z., Liu, W., Gao, H., Huang, H., Ma, Z., Jiang, X., Chen, Y., et al.: Grounding DINO 1.5: Advance the "edge" of open-set object detection. arXiv preprint arXiv:2405.10300 (2024)

    work page arXiv 2024

  43. [43]

    Grounded SAM: Assembling Open-World Models for Diverse Visual Tasks

    Ren, T., Liu, S., Zeng, A., Lin, J., Li, K., Cao, H., Chen, J., Huang, X., Chen, Y., Yan, F., et al.: Grounded SAM: Assembling open-world models for diverse visual tasks. arXiv preprint arXiv:2401.14159 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  44. [44]

    In: CVPR

    Schönberger, J.L., Frahm, J.M.: Structure-from-motion revisited. In: CVPR. pp. 4104–4113 (2016)

    2016

  45. [45]

    Food Chemistry424, 136309 (2023)

    Shao, W., Min, W., Hou, S., Luo, M., Li, T., Zheng, Y., Jiang, S.: Vision-based food nutrition estimation via RGB-D fusion network. Food Chemistry424, 136309 (2023)

    2023

  46. [46]

    DINOv3

    Siméoni, O., Vo, H.V., Seitzer, M., Baldassarre, F., Oquab, M., Jose, C., Khali- dov, V., Szafraniec, M., Yi, S., Ramamonjisoa, M., et al.: Dinov3. arXiv preprint arXiv:2508.10104 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  47. [47]

    JMIR mHealth and uHealth9(1), e24467 (2021)

    Vasiloglou, M.F., Van der Horst, K., Stathopoulou, T., Jaeggi, M.P., Tedde, G.S., Lu, Y., Mougiakakou, S.: The human factor in automated image-based nutrition apps: Analysis of common mistakes using the gofood lite app. JMIR mHealth and uHealth9(1), e24467 (2021)

    2021

  48. [48]

    In: CVPRW

    Vinod, G., He, J., Shao, Z., Zhu, F.: Food portion estimation via 3d object scaling. In: CVPRW. pp. 3741–3749 (2024)

    2024

  49. [49]

    In: CVPR

    Wang, J., Chen, M., Karaev, N., Vedaldi, A., Rupprecht, C., Novotny, D.: Vggt: Visual geometry grounded transformer. In: CVPR. pp. 5294–5306 (2025)

    2025

  50. [50]

    In: ACM TOG

    Wang, Y., Sun, Y., Liu, Z., Sarma, S.E., Bronstein, M.M., Solomon, J.M.: Dynamic graph cnn for learning on point clouds. In: ACM TOG. ACM (2019)

    2019

  51. [51]

    In: CVPR

    Xiang, J., Lv, Z., Xu, S., Deng, Y., Wang, R., Zhang, B., Chen, D., Tong, X., Yang, J.: Structured 3D latents for scalable and versatile 3D generation. In: CVPR. pp. 21469–21480 (2025)

    2025

  52. [52]

    Xu, K., Hu, W., Leskovec, J., Jegelka, S.: How powerful are graph neural networks? In: ICLR (2019)

    2019

  53. [53]

    Yu, T., Zheng, Z., Guo, K., Liu, P., Dai, Q., Liu, Y.: Function4D: Real-time human volumetriccapturefromverysparseconsumerRGBDsensors.In:CVPR.pp.5742– 5752 (2021)

    2021

  54. [54]

    Coral Reefs41(4), 973–985 (2022)

    Yudelman, E.A., Slowey, N.C.: Coral extension rate analysis using computed axial tomography. Coral Reefs41(4), 973–985 (2022)

    2022

  55. [55]

    In: Proceedings of the 2010 IEEE 36th Annual Northeast Bioengineering Con- ference (NEBEC)

    Yue, Y., Jia, W., Fernstrom, J.D., Sclabassi, R.J., Fernstrom, M.H., Yao, N., Sun, M.: Food volume estimation using a circular reference in image-based dietary stud- ies. In: Proceedings of the 2010 IEEE 36th Annual Northeast Bioengineering Con- ference (NEBEC). pp. 1–2 (2010) 1

    2010

  56. [56]

    In: ICCV

    Zhao, H., Jiang, L., Jia, J., Torr, P.H., Koltun, V.: Point transformer. In: ICCV. pp. 16259–16268 (2021)

    2021

  57. [57]

    Hunyuan3D 2.0: Scaling Diffusion Models for High Resolution Textured 3D Assets Generation

    Zhao, Z., Lai, Z., Lin, Q., Zhao, Y., Liu, H., Yang, S., Feng, Y., Yang, M., Zhang, S., Yang, X., et al.: Hunyuan3d 2.0: Scaling diffusion models for high resolution textured 3d assets generation. arXiv preprint arXiv:2501.12202 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  58. [58]

    Ziqiang, Z., Yaofeng, X., Haixin, L., Zhibin, Y., Yeung, S.K.: CoralVOS: Dataset and benchmark for coral video segmentation. arXiv preprint arXiv:2310.01946 (2023) A Datasets Preparation Details In this section, we describe the data generation and preprocessing pipelines used to create our training and evaluation datasets: Objaverse [7] (for pre-training)...

    work page arXiv 2023