MVImgNet2.0: A Larger-scale Dataset of Multi-view Images

Haolin Liu; Hongjie Liao; Lingteng Qiu; Luyue Shi; Shuguang Cui; Weihao Yuan; Xiaodong Gu; Xiaoguang Han; Yushuang Wu; Zilong Dong

REVIEW 2 major objections 5 minor 3 cited by

MVImgNet2.0: A Larger-scale Dataset of Multi-view Images

T0 review · 2 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read MVImgNet2.0 expands multi-view real-object imagery to 520k objects in 515 categories and shows that the larger scale, 360-degree coverage, and higher-accuracy masks, camera poses, and point clouds improve large 3D reconstruction models.

desk verdict A genuinely large and useful dataset expansion whose scale and mask/pose improvements are real, but the headline point-cloud-quality result is weakened by an evaluation target produced by the same pipeline being compared. read the letter →

arxiv 2412.01430 v1 pith:7NBLU45V submitted 2024-12-02 cs.CV cs.AIcs.GR

Xiaoguang Han , Yushuang Wu , Luyue Shi , Haolin Liu , Hongjie Liao , Lingteng Qiu , Weihao Yuan , Xiaodong Gu

show 2 more authors

Zilong Dong Shuguang Cui

This is my paper · ORCID

classification cs.CVcs.AIcs.GR

keywords 3Dobjectdatasetmulti-viewimagesreconstructionimage-basedmodelinglarge-scalepointcloudcameraposeestimationsegmentation

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

MVImgNet2.0 expands the existing MVImgNet dataset from about 220,000 to roughly 520,000 real-world objects across 515 categories, adding 300,000 new crowdsourced object videos, most captured in full 360-degree orbits, and re-annotating everything with more accurate foreground masks, camera poses, and dense point clouds. The paper's central claim is that this larger, higher-quality real-image dataset supplies a stronger 3D prior for category-agnostic object reconstruction, and it supports the claim by training three large reconstruction models—LGM, LRM, and TriplaneGaussian—on the new data, the old data, and synthetic data. The new data alone improves reconstruction quality over the old data despite being evaluated as a standalone training set, and combining both datasets improves it further; per-scene reconstructions also improve when the new camera poses are used. The work matters because 3D deep learning lacks a real-world dataset at the scale of 2D benchmarks like ImageNet, and a resource of this size and quality could let reconstruction models learn from the physical world instead of synthetic CAD models.

What carries the argument

The load-bearing machinery is the dataset itself together with its annotation pipeline. Raw data consists of crowdsourced object-centric videos shot with phone cameras orbiting the object, most covering a full 360-degree view. Annotations are produced by three upgraded components: Pixel-Perfect Structure-from-Motion (PixSfM) for camera poses, which improves keypoint localization and bundle adjustment using dense feature metric refinement; a detection-segmentation-tracking pipeline that combines Grounding-DINO (open-set detection), SAM (segmentation), and DeAOT (video object tracking) to generate foreground masks; and Neural-Angelo/Instant-Angelo neural surface reconstruction with multi-resolution hash grids for dense point clouds. In the experiments, the large reconstruction models LGM, LRM, and TriplaneGaussian act as the measuring instruments: retraining them on MV2-Data versus MV1-Data versus synthetic Objaverse isolates the contribution of the new dataset, and the LGM-tiny ablations systematically vary data scale, category count, and 360-degree view ratio to attribute the observed gains.

What would settle it

Check whether the 20 held-out test categories are disjoint from the 238 categories in MVImgNet1.0; if any overlap exists, re-run Tables 3 and 4 using test categories provably absent from both training sets. As an additional check, train LGM on equal-size, category-matched subsets of MV1-Data and MV2-Data; if the PSNR gap between them disappears or reverses, the claim that the new data's higher quality drives the improvement would not hold.

Watch

Extended reading notes

Core claim

MVImgNet2.0 is a dataset of roughly 520,000 real-life objects in 515 categories, formed by combining the original 220,000-object MVImgNet with 300,000 newly collected object videos that cover 347 classes, 277 of them new. Compared with its predecessor, most of the new videos circle the object through a full 360-degree view; foreground masks come from a detection-segmentation-tracking pipeline built on Grounding-DINO, SAM, and DeAOT rather than from CarveKit; camera poses come from Pixel-Perfect SfM, which refines keypoints and bundle adjustment with dense features; and dense point clouds come from a Neural-Angelo-based reconstruction with multi-resolution hash encoding. The paper's central discovery is that training large reconstruction models on this data improves both reconstruction quality and generalization: LGM and LRM trained on the new data alone outperform the same models trained on the old data, and training on both datasets improves them further; TriplaneGaussian trained with the 360-degree views and the new point-cloud supervision achieves markedly higher PSNR and lower Chamfer distance; and controlled ablations show that data scale, category breadth, 360-degree view ratio, and MV2-style annotation quality each separately raise reconstruction PSNR.

Load-bearing premise

The experiments that attribute the performance gains to MVImgNet2.0's higher quality assume that the 20 held-out test categories are equally unfamiliar to models trained on the old MVImgNet data and on the new data, so that neither training set has already memorized the test categories; if those test categories overlap with the original 238 categories, the reported advantage of MV2-Data over MV1-Data would be biased in favor of the new data.

Editorial extensions

If this is right

If the claims hold, researchers gain a real-image multi-view dataset at about half ImageNet's scale, allowing 3D reconstruction and generation models to be trained or fine-tuned with less dependence on synthetic CAD data.
The improved camera-pose annotations alone produce large per-scene gains—about 1.1 dB for Instant-NGP and 5.8 dB for 3D Gaussian Splatting—so downstream users should get better novel-view synthesis and radiance-field fitting from the same videos.
The 360-degree coverage and denser point-cloud supervision should make reconstructed shapes more complete than the 180-degree MVImgNet views, which leave backsides and occluded regions poorly constrained.
The scaling curves in the ablation study show continued improvement as added objects grow to 140k, suggesting that further collection along the same pipeline would keep yielding gains rather than saturating.
The combined MVImgNet1.0+2.0 corpus, with 520k objects across 515 categories, becomes a more credible real-world counterpart to large synthetic datasets like Objaverse for training category-agnostic reconstruction models.

Reading between the lines

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

The paper never states whether the 20 held-out test categories overlap with the original 238 MVImgNet categories; if they do, the MV2-Data advantage in Tables 3 and 4 could partly reflect category leakage rather than annotation quality. This is an editorial caution, not a claim made in the paper.
The large per-scene pose effect (about 5.8 dB for 3DGS) hints that camera-pose error, more than mask accuracy or point-cloud density, is the binding constraint for high-frequency radiance-field training; the paper's design does not fully separate these three annotation improvements.
Because the earlier MVImgNet already proved useful for view-consistent understanding, multi-view diffusion, and video generation, MVImgNet2.0's larger scale and broader categories will likely benefit those tasks too, but the paper only tests reconstruction, so those transfers remain unverified.
A testable extension would be to train reconstruction models on mixed Objaverse + MVImgNet2.0 data and measure whether the real-to-synthetic domain gap narrows relative to training on either alone; the paper's comparisons treat the two sources separately rather than jointly.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Bottom line: this is a substantial dataset contribution—300k new objects, 277 new categories, mostly 360-degree captures—and the mask and pose improvements are backed by reasonable evidence. The point-cloud quality claim is the soft spot. Table 4's Chamfer-Distance target is generated by the same MV2 annotation pipeline (PixSfM + Instant-Angelo) used as training supervision, so a model trained on MV2 point clouds has an unfair advantage in matching those targets. Manual cleaning does not remove the shared systematic bias. The CD drop from 0.89 to 0.36 is therefore partly an alignment artifact, not proof that MV2 point clouds are absolutely more accurate. The rendering metrics in the same table are less affected, so the broader claim that MV2 data helps large reconstruction models still stands; it is specifically the "higher-quality point clouds improve shape reconstruction" conclusion that is not established by this experiment. A cleaner evaluation would compare against an independent reconstruction (e.g., laser scans or a different photogrammetry pipeline) or at least caveat the result.

The other issues are minor. There are no error bars or significance tests on any of the tables; with 50 per-scene objects and 1k test samples, reporting variance would be easy and should be added. The held-out test categories are said to be 'unseen in training' but not explicitly stated to be disjoint from MVImgNet1.0's 238 categories; if they overlap, that biases the comparison against MV2, so it is a clarification rather than a fatal flaw.

What the paper does well: the dataset itself is the contribution. Mask segmentation improvements are validated on external datasets (ECSSD, DAVIS) plus a manual subset, which is the right kind of evidence. Per-scene reconstruction with PixSfM poses shows a clear improvement, especially for 3DGS (5.8 dB). Training LGM, LRM, and TriplaneGaussian on MV2 data yields consistent rendering gains, so the practical value of the data is believable. The limitations section is honest—excluded categories, annotation refinement, and simple camera trajectories are all acknowledged.

Who is this for? Anyone working on 3D reconstruction, novel view synthesis, or multi-view learning; the released dataset alone justifies the paper. It deserves a serious referee, not a desk reject. I'd send it to review with a request to add error bars, clarify the test category overlap, and either fix or substantially temper the CD-based point-cloud claim.

Referee Report

2 major / 5 minor

Summary. The paper introduces MVImgNet2.0, a multi-view image dataset that extends MVImgNet from roughly 220k objects in 238 classes to about 520k objects in 515 classes, with newly collected videos mostly covering 360-degree views. It also updates the annotation pipeline: masks via a detection-segmentation-tracking approach, camera poses via PixSfM, and dense point clouds via Instant-Angelo/Neural-Angelo. The authors claim these new features yield higher-quality annotations and demonstrate that training large reconstruction models (LGM, LRM, TriplaneGaussian) on the new data improves rendering quality and Chamfer distance compared with training on Objaverse or MVImgNet1.0 data. Per-scene reconstruction with INGP and 3DGS also improves when the MVImgNet2.0 camera poses are used. The dataset, reconstructed point clouds, and annotation code are to be released publicly.

Significance. If the claims hold, MVImgNet2.0 is a substantial community resource: no existing real-world multi-view object dataset combines this scale, category coverage, and 360-degree coverage, and the consistent improvements across three large reconstruction models in Tables 3 and 4 are encouraging. The controlled ablations in Figure 9, isolating data scale, category range, and view range, are a useful addition, and the mask-quality evaluation in the supplementary material includes external datasets (ECSSD, DAVIS) plus a manually annotated subset, which is more than many dataset papers provide. However, the headline shape-quality claim rests on a Chamfer-distance evaluation whose test target is generated by the same reconstruction pipeline used to create the training supervision; that confound must be addressed before the higher-quality point-cloud claim is accepted.

major comments (2)

[Table 4; Sections 3.2 and 4.1] The Chamfer-Distance comparison in Table 4 is not an independent evaluation of shape quality. Section 4.1 states that the test shape ground truths are "high-quality dense point cloud reconstructions with manual cleaning by annotators," and Section 3.2 describes these as outputs of the MVImgNet2.0 pipeline (PixSfM poses plus Instant-Angelo/Neural-Angelo dense reconstruction). Therefore, the row trained with MV2-Anno point-cloud supervision is trained to imitate the same reconstruction procedure that generated the test target. The reported drop in CD from 0.89 to 0.36, and even the 0.40 for the Objaverse-trained model, can partly reflect alignment of systematic biases such as smoothness, completeness, or hallucinated regions rather than improved absolute geometry. Manual cleaning edits the same reconstructions and does not supply an independent measurement. Because feature (iv) of the paper is one of the four headline claims and Table 4 is its main quantitative support, the shape-quality conclusion is not established by the current experiment. Please add an external shape ground truth, for example a small set of objects measured by laser scanning or high-fidelity RGB-D scanning, and report CD against that target; alternatively, rephrase the claim as consistency with the MVImgNet2.0 annotation style and support it with rendering metrics alone.
[Tables 2-4; Section 4.1] The paper reports only single mean values over the 50 per-scene objects and the 1k test samples, with no error bars, confidence intervals, or multi-seed variance. In Table 3, the reported gains of about 0.3-0.5 dB in PSNR are small enough that without variance information it is difficult to rule out run-to-run or test-set sampling variation, even though the direction of the effect is consistent across models. Please report at least bootstrap confidence intervals over the test set and, ideally, standard deviations over a few independent training runs, for the key comparisons in Tables 3 and 4 and Figure 9.

minor comments (5)

[Section 4.1] The text says the 20 held-out test categories are "unseen in training" but does not state whether they overlap with the 238 MVImgNet1.0 categories or are among the 277 new categories. If they overlap, the MV1-Data baseline has an advantage, which would make the reported MV2 wins conservative; if they are disjoint, the comparison is more directly interpretable. Please state the overlap explicitly.
[Table 2] The last column header reads "LIPIS"; this should be "LPIPS".
[Supplementary Table R.1] The dataset name is written "ECCSD"; the correct name is ECSSD.
[Section 3.3] The text says MVImgNet2.0 includes 300k videos and 347 object classes, while the abstract emphasizes the combined 520k objects and 515 categories; the relationship between the newly added portion and the combined total should be stated in one place so that the two numbers do not appear inconsistent.
[Figure 9] The caption says the figure analyzes three factors, but the figure contains four panels, with panel (d) varying the ratio of MV2-Data at constant total scale; please update the caption to describe all four panels.

Circularity Check

1 steps flagged · score 6.0 of 10

Table 4's Chamfer-Distance evidence for higher-quality MV2-Anno point clouds is circular: the test shape ground truth is produced by the same Neural-Angelo pipeline used as MV2-Anno training supervision.

fitted input called prediction [Section 4.1 (Evaluation metrics) and Table 4; Section 3.2 (Dense reconstruction)]
"we also provide high-quality dense point cloud reconstructions with manual cleaning by annotators for each sample in the test set as their shape ground truths. ... we advance the point cloud reconstruction approach based on a neural surface reconstruction method, Neural-Angelo ... Given the Neural-Angelo reconstruction outputs, similar to MVImgNet, we also manually clean the point clouds to delete the objects with obvious noisy, extremely sparse reconstructions, or backgrounds."

The test set's shape ground truths are the same Neural-Angelo/Instant-Angelo outputs that define the MV2-Anno point-cloud supervision used in training. Table 4 compares TriplaneGaussian trained with MV1-Anno versus MV2-Anno point-cloud supervision and reports Chamfer Distance against these MV2-Anno test targets. A model trained to imitate MV2-Anno point clouds will match MV2-Anno-specific biases (smoothness, hallucinated regions, completeness behavior) and thereby achieve lower CD even if its absolute geometric accuracy is no better; manual cleaning edits the same reconstruction rather than independently measuring the object.

full rationale

The paper's core dataset-construction claims (scale, category expansion, 360-degree capture) are not circular: they are factual descriptions of collected data. The mask-quality evaluation in SupMat Tab. R.1 uses external datasets (ECSSD, DAVIS) plus a manually annotated subset, which is independent support. The per-scene pose comparison (Tab. 2) directly rewards better camera poses through novel-view PSNR, which is a legitimate validation. The LGM/LRM rendering comparisons (Tab. 3) evaluate against 2D novel-view images, not against the same point-cloud pipeline being promoted, so they are not circular; the held-out-category overlap concern, if anything, would make MV2-Data's win conservative. The one genuinely circular step is the Chamfer-Distance evaluation in Table 4: the test-set shape ground truths are explicitly described as high-quality dense point cloud reconstructions with manual cleaning, and Section 3.2 shows those reconstructions are produced by the same Neural-Angelo/Instant-Angelo pipeline that constitutes the MV2-Anno supervision. Thus the CD comparison measures how well a model imitates the MV2-Anno output distribution, not independent geometric quality. This makes the point-cloud-quality claim partially circular, while the rendering-based evidence and the dataset-scale contribution retain independent content.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The central claims rest on the accuracy of the annotation pipeline (poses, masks, point clouds) and on the fairness of the benchmark design. No free parameters are fitted in this paper; the key assumptions are about tool effectiveness and experiment design rather than about derived constants.

assumptions (5)

domain assumption The camera poses estimated by PixSfM are more accurate than classical SfM, particularly for low-texture objects.
Invoked in Section 3.2 (Sparse reconstruction) and supported only indirectly by improved per-scene reconstruction (Table 2); no direct pose-error measurement is reported.
domain assumption The Grounding-DINO + SAM + DeAOT pipeline produces more accurate foreground masks than CarveKit.
Invoked in Section 3.2; quantitatively evaluated only on a 500-frame subset of MVImgNet, external datasets (ECSSD, DAVIS) in SupMat Table R.1, not on the new data itself.
domain assumption The manually cleaned dense point clouds are accurate enough to serve as shape ground truth.
Used as supervision for TriplaneGaussian and as test set ground truth (Sections 4.1 and 4.3); no inter-annotator agreement or independent accuracy metric is given.
ad hoc to paper The 20 held-out test categories are equally novel to models trained on MV1-Data and MV2-Data.
Section 4.1 states the test set comes from held-out categories in MVImgNet2.0, but does not clarify whether those categories overlap with the 238 MVImgNet1.0 categories; if they overlap, the MV1 vs MV2 comparison is biased.
domain assumption Default hyperparameters from official implementations give fair comparisons across data sources.
Section 4.1 (Implementation details) states all hyper-parameters follow recommended/default settings, which assumes no data-specific tuning confounds the comparisons.

how reviews work

0 comments

Cite this review

Pith. "Pith review of MVImgNet2.0: A Larger-scale Dataset of Multi-view Images." pith.science (2026). https://pith.science/paper/7NBLU45V

@misc{pith2026241201430,
  author       = {Pith},
  title        = {Pith review of: MVImgNet2.0: A Larger-scale Dataset of Multi-view Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7NBLU45V}},
  note         = {Machine review of arXiv:2412.01430}
}

read the original abstract

MVImgNet is a large-scale dataset that contains multi-view images of ~220k real-world objects in 238 classes. As a counterpart of ImageNet, it introduces 3D visual signals via multi-view shooting, making a soft bridge between 2D and 3D vision. This paper constructs the MVImgNet2.0 dataset that expands MVImgNet into a total of ~520k objects and 515 categories, which derives a 3D dataset with a larger scale that is more comparable to ones in the 2D domain. In addition to the expanded dataset scale and category range, MVImgNet2.0 is of a higher quality than MVImgNet owing to four new features: (i) most shoots capture 360-degree views of the objects, which can support the learning of object reconstruction with completeness; (ii) the segmentation manner is advanced to produce foreground object masks of higher accuracy; (iii) a more powerful structure-from-motion method is adopted to derive the camera pose for each frame of a lower estimation error; (iv) higher-quality dense point clouds are reconstructed via advanced methods for objects captured in 360-degree views, which can serve for downstream applications. Extensive experiments confirm the value of the proposed MVImgNet2.0 in boosting the performance of large 3D reconstruction models. MVImgNet2.0 will be public at luyues.github.io/mvimgnet2, including multi-view images of all 520k objects, the reconstructed high-quality point clouds, and data annotation codes, hoping to inspire the broader vision community.

Figures

Figures reproduced from arXiv: 2412.01430 by the authors.

**Figure 1.** We introduce MVImgnet2.0, a larger-scale dataset of multi-view images, which enjoys 3D-aware signals from multi-view consistency. MVImgnet2.0 expands its last version into a total of 520k objects from 515 categories, and also provides higher-quality annotations. The extremely rich geometry and texture information in real-world objects leads to MVImgNet2.0’s great potential in supporting large-scale learning in the 3… view at source ↗

**Figure 2.** MVImgNet2.0 data visualization. Objects in MVImgNet2.0 are in a wide range. For each object, we visualize the estimated camera poses and then [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** The data acquisition and annotation pipeline in MVImgNet2.0. One [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (8 more)

**Figure 4.** Figure 4: The sparse reconstruction results comparison between using the orig [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: The comparison of the foreground object segmentation results be [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

**Figure 6.** Figure 6: The dense point cloud reconstruction results comparison between [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]

**Figure 8.** Figure 8: Qualitative results of TriplaneGaussian trained on different view [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]

**Figure 7.** Figure 7: Qualitative results of LGM and LRM trained on different data from [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

**Figure 9.** Figure 9: The experiments for analyzing the effects of three factors in training [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]

**Figure 10.** Figure 10: Visualization of more multi-view data from different object categories in MVImgNet2.0. [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]

**Figure 11.** Figure 11: Visualization of more reconstructed point cloud annotations from different object categories in MVImgNet2.0. [PITH_FULL_IMAGE:figures/full_fig_p011_11.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

TRACE: High-Fidelity 3D Scene Editing via Tangible Reconstruction and Geometry-Aligned Contextual Video Masking
cs.CV 2026-04 conditional novelty 6.0 of 10

TRACE anchors a video-diffusion editor to 3D meshes to perform consistent part-level edits on 3D Gaussian scenes in about 10 minutes per edit.
iMontage: Unified, Versatile, Highly Dynamic Many-to-many Image Generation
cs.CV 2025-11 conditional novelty 6.0 of 10

iMontage repurposes a pretrained video diffusion model to generate coherent yet highly dynamic image sets from arbitrary numbers of input images.
Droplet3D: Commonsense Priors from Videos Facilitate 3D Generation
cs.CV 2025-08 conditional novelty 6.0 of 10

A video diffusion backbone fine-tuned on 4M densely captioned 360-degree renderings generates spatially consistent multi-view images for 3D assets from image plus detailed text input.

Reference graph

Works this paper leans on

8 extracted references · 2 canonical work pages · cited by 3 Pith papers

[6]

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

MVImgNet: A Large-scale Dataset of Multi-view Images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . Kai Zhang, Sai Bi, Hao Tan, Yuanbo Xiangli, Nanxuan Zhao, Kalyan Sunkavalli, and Zexiang Xu. 2024. GS-LRM: Large Reconstruction Model for 3D Gaussian Splatting. arXiv preprint arXiv:2404.19702 (2024). Richard ...

arXiv 2024
[7]

CaesarNeRF: Calibrated Semantic Representation for Few-shot Generalizable Neural Rendering

CaesarNeRF: Calibrated Semantic Representation for Few-shot Generalizable Neural Rendering. arXiv preprint arXiv:2311.15510 (2023). Zi-Xin Zou, Zhipeng Yu, Yuan-Chen Guo, Yangguang Li, Ding Liang, Yan-Pei Cao, and Song-Hai Zhang. 2023. Triplane meets gaussian splatting: Fast and generalizable single-view 3d reconstruction with transformers. arXiv preprint...

work page Pith review arXiv 2023
[500]

Methods ECCSD DAVIS MV1-500 MV1-Anno 0.143 0.195 0.243 MV2-Anno (ours) 0.103 0.143 0.172 B MORE EXPERIMENTS Mask annotation quality

with ground-truth object masks. Methods ECCSD DAVIS MV1-500 MV1-Anno 0.143 0.195 0.243 MV2-Anno (ours) 0.103 0.143 0.172 B MORE EXPERIMENTS Mask annotation quality. In the annotation process, we adopt a detection-segmentation-tracking pipeline to generate the fore- ground object mask in each view. To better demonstrate the supe- riority of the used segmen...

work page 2015
[2019]

In Proceedings of the IEEE/CVF International Conference on Computer Vision

C3dpo: Canonical 3d pose networks for non-rigid structure from motion. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 7688–7697. Evangelos Ntavelis, Aliaksandr Siarohin, Kyle Olszewski, Chaoyang Wang, Luc V Gool, and Sergey Tulyakov. 2023. Autodecoding latent 3d diffusion models. Advances in Neural Information Processing Syste...

arXiv 2023
[2020]

International Journal of Computer Vision (IJCV) 128, 7 (2020), 1956–1981

The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale. International Journal of Computer Vision (IJCV) 128, 7 (2020), 1956–1981. ACM Trans. Graph., Vol. 43, No. 6, Article . Publication date: December 2024. MVImgNet2.0: A Larger-scale Dataset of Multi-view Images • 13 Han-Hung Lee, Yiming Zh...

arXiv 2020
[2022]

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

High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 10684–10695. Joseph Roth, Yiying Tong, and Xiaoming Liu. 2016. Adaptive 3D face reconstruction from unconstrained photo collections. In Proceedings of the IEEE conference on computer vision and pattern recogn...

arXiv 2022
[2023]

arXiv preprint arXiv:2307.09288 (2023)

Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288 (2023). Shubham Tulsiani, Tinghui Zhou, Alexei A Efros, and Jitendra Malik. 2017. Multi- view supervision for single-view reconstruction via differentiable ray consistency. In Proceedings of the IEEE conference on computer vision and pattern recognition . 2626–2634. Mikae...

arXiv 2023
[2024]

arXiv preprint arXiv:2403.19632 (2024)

GauStudio: A Modular Framework for 3D Gaussian Splatting and Beyond. arXiv preprint arXiv:2403.19632 (2024). Alex Yu, Vickie Ye, Matthew Tancik, and Angjoo Kanazawa. 2021. pixelnerf: Neural radiance fields from one or few images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 4578–4587. Xianggang Yu, Mutian Xu, Yida...

arXiv 2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [6]

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

MVImgNet: A Large-scale Dataset of Multi-view Images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . Kai Zhang, Sai Bi, Hao Tan, Yuanbo Xiangli, Nanxuan Zhao, Kalyan Sunkavalli, and Zexiang Xu. 2024. GS-LRM: Large Reconstruction Model for 3D Gaussian Splatting. arXiv preprint arXiv:2404.19702 (2024). Richard ...

arXiv 2024

[2] [7]

CaesarNeRF: Calibrated Semantic Representation for Few-shot Generalizable Neural Rendering

CaesarNeRF: Calibrated Semantic Representation for Few-shot Generalizable Neural Rendering. arXiv preprint arXiv:2311.15510 (2023). Zi-Xin Zou, Zhipeng Yu, Yuan-Chen Guo, Yangguang Li, Ding Liang, Yan-Pei Cao, and Song-Hai Zhang. 2023. Triplane meets gaussian splatting: Fast and generalizable single-view 3d reconstruction with transformers. arXiv preprint...

work page Pith review arXiv 2023

[3] [500]

Methods ECCSD DAVIS MV1-500 MV1-Anno 0.143 0.195 0.243 MV2-Anno (ours) 0.103 0.143 0.172 B MORE EXPERIMENTS Mask annotation quality

with ground-truth object masks. Methods ECCSD DAVIS MV1-500 MV1-Anno 0.143 0.195 0.243 MV2-Anno (ours) 0.103 0.143 0.172 B MORE EXPERIMENTS Mask annotation quality. In the annotation process, we adopt a detection-segmentation-tracking pipeline to generate the fore- ground object mask in each view. To better demonstrate the supe- riority of the used segmen...

work page 2015

[4] [2019]

In Proceedings of the IEEE/CVF International Conference on Computer Vision

C3dpo: Canonical 3d pose networks for non-rigid structure from motion. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 7688–7697. Evangelos Ntavelis, Aliaksandr Siarohin, Kyle Olszewski, Chaoyang Wang, Luc V Gool, and Sergey Tulyakov. 2023. Autodecoding latent 3d diffusion models. Advances in Neural Information Processing Syste...

arXiv 2023

[5] [2020]

International Journal of Computer Vision (IJCV) 128, 7 (2020), 1956–1981

The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale. International Journal of Computer Vision (IJCV) 128, 7 (2020), 1956–1981. ACM Trans. Graph., Vol. 43, No. 6, Article . Publication date: December 2024. MVImgNet2.0: A Larger-scale Dataset of Multi-view Images • 13 Han-Hung Lee, Yiming Zh...

arXiv 2020

[6] [2022]

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

High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 10684–10695. Joseph Roth, Yiying Tong, and Xiaoming Liu. 2016. Adaptive 3D face reconstruction from unconstrained photo collections. In Proceedings of the IEEE conference on computer vision and pattern recogn...

arXiv 2022

[7] [2023]

arXiv preprint arXiv:2307.09288 (2023)

Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288 (2023). Shubham Tulsiani, Tinghui Zhou, Alexei A Efros, and Jitendra Malik. 2017. Multi- view supervision for single-view reconstruction via differentiable ray consistency. In Proceedings of the IEEE conference on computer vision and pattern recognition . 2626–2634. Mikae...

arXiv 2023

[8] [2024]

arXiv preprint arXiv:2403.19632 (2024)

GauStudio: A Modular Framework for 3D Gaussian Splatting and Beyond. arXiv preprint arXiv:2403.19632 (2024). Alex Yu, Vickie Ye, Matthew Tancik, and Angjoo Kanazawa. 2021. pixelnerf: Neural radiance fields from one or few images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 4578–4587. Xianggang Yu, Mutian Xu, Yida...

arXiv 2024