Pith. sign in

REVIEW 3 major objections 5 minor 18 references

Temporally Grounded Compositional Camera Motion Understanding via Geometric Knowledge Distillation

T0 review · 3 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 Camera-motion model keeps 3D teacher's accuracy after dropping it

desk verdict A genuinely useful benchmark and a well-controlled distillation comparison, but the missing inter-annotator agreement and single-run results leave the anchor claim softer than the tables suggest. read the letter →

arxiv 2608.10932 v1 pith:P7J2XQ2M submitted 2026-08-11 cs.CV cs.AI

Dazhao Du , Shiyan Du , Jian Liu , Yongjian Yu , Bohai Gu , Tao Han , Hualuo Liu , Eric Liu
show 3 more authors
Yujia Zhang Xi Chen Song Guo
This is my paper · ORCID
classification cs.CVcs.AI
keywords cameramotiontemporalgroundingcompositionalrecognitionmultimodallargelanguagemodelknowledgedistillation3Dgeometryvideounderstandingbenchmark
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 camera-motion understanding should be treated as a temporally grounded, compositional recognition problem: a model must split a shot into motion-consistent intervals and, for each interval, list every simultaneous movement (pan, dolly, tilt, zoom, and so on). To support this, it introduces CamChoreo, a benchmark of 4,229 real single-shot clips with expert-annotated segments, 20 direction-aware labels, and compound motion in 44.2% of segments. It then shows that current multimodal large language models score poorly because their vision encoders are semantic, not geometric. The central claim is that distilling a frozen 3D geometry teacher's per-frame camera tokens into lightweight student tokens (CamDistill) matches the accuracy of directly injecting the teacher's features (CamInject) while removing the 3D model at inference. A sympathetic reader should care because, if true, expensive geometric supervision can be compressed into a cheap per-frame stream that is free at test time.

What carries the argument

The load-bearing mechanism is the Geometry-aware Camera Token Extractor (GCTE), a lightweight branch attached to the frozen vision encoder. GCTE predicts one camera token per frame using alternating frame-wise cross-attention (each camera token reads only its own frame's intermediate visual features) and global camera self-attention (tokens exchange temporal context). A cosine-distance distillation loss aligns the per-frame student tokens with the teacher's cached 2048-dimensional camera tokens. At inference the teacher is removed, leaving a compact geometry-aware stream inserted before each frame's visual tokens.

What would settle it

Re-annotate a random subset of CamChoreo (say 300 clips) with a fresh expert panel following the same protocol and measure inter-annotator agreement on boundaries and label sets, especially for Dolly/Zoom, Pan/Truck, and compound labels. If per-label agreement falls below a high threshold, the benchmark's difficulty numbers and the distillation parity claim lose their footing. A second check: train a linear probe on CamDistill's student tokens to predict the teacher's camera pose; if the probe achieves low accuracy, the tokens may not carry the geometry the paper attributes to them.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the benefit of an expensive 3D geometry model for camera-motion recognition can be preserved without running it at test time. Using VGGT-Ω's per-frame camera token as a distillation target, CamDistill trains a Geometry-aware Camera Token Extractor (GCTE) on frozen intermediate vision features, then discards the teacher. On CamChoreo, CamDistill-4B reaches 67.5 frame-level micro F1, exactly matching CamInject-4B's 67.5, with SegDet@0.5 within 0.4; the 8B version trails direct injection by 0.5 micro F1. Inference latency stays at 10.2 seconds per clip versus 10.1 seconds for the base model and 16.0 seconds for CamInject. The distilled tokens also transfer to external clip-level benchmarks (CameraBench and CMVQA), improving on both the base model and a CameraBench-tuned specialist.

Load-bearing premise

The load-bearing premise is that the five annotators' temporal segments and 20-label sets are reliable ground truth; no inter-annotator agreement statistic is reported, so if hard distinctions like Dolly versus Zoom or Pan versus Truck are inconsistent across annotators, the benchmark rankings and the CamDistill–CamInject parity would rest on noisy labels.

Editorial extensions

If this is right

  • If the parity holds, video multimodal LLMs can be trained once with a 3D teacher and shipped without it, retaining most of the geometric benefit at near-base inference cost.
  • Temporally grounded, compositional camera labels become a practical target for large-scale supervised training, since the benchmark's annotations support precise segment-level evaluation.
  • Rare geometry-dependent classes (Zoom, Roll, Focus Shift) gain more from distillation than frequent ones, so geometry supervision should be a standard component for camera-aware video models.
  • The same token-distillation scheme could generalize to other geometry-heavy video tasks, such as object-motion or depth-based reasoning, wherever a teacher produces per-frame tokens.
  • External transfer results imply the distilled representation carries reusable camera cues, not just benchmark-format memorization.

Reading between the lines

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

  • The paper does not report inter-annotator agreement; an independent re-annotation study could reveal how much of the 20-point gap over closed-source models is label noise versus model capability.
  • If the student tokens truly capture pose-associated geometry, they should be decodable into camera pose by a linear probe; the paper does not test this, but such a probe would directly test whether distillation transfers geometry rather than only task-specific cues.
  • Because the teacher is removed at inference, the same approach could be used with an even larger or multi-view teacher at training time only, pushing accuracy without affecting inference cost.
  • The benchmark restricts to single-shot clips; extending to edited multi-shot video would test whether temporally grounded compositional labels survive cut boundaries and narrative structure.
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 / 5 minor

Summary. The paper formulates camera-motion understanding as temporally grounded compositional recognition, where a video must be segmented into motion-consistent intervals and each interval labeled with the complete set of active direction-aware movements. It introduces CAMCHOREO, a benchmark of 4,229 real single-shot clips from YouTube with 8,591 segments and 14,258 motion instances over a 20-label taxonomy, annotated by five experts. It also proposes CAMDISTILL, which attaches a lightweight Geometry-aware Camera Token Extractor (GCTE) to a frozen vision encoder, trains it to regress VGGT-Omega camera tokens via a cosine-distance distillation loss, and removes the teacher at inference. The central empirical claim is that CAMDISTILL matches CAMINJECT, direct injection of teacher features, on CAMCHOREO (67.5 vs 67.5 micro F1 on the 4B backbone; 67.8 vs 68.3 on 8B) while adding only about 0.1 s per clip and 1.8 GB memory over the base model. The paper also reports external transfer to CameraBench and CMVQA and extensive ablations.

Significance. If the central claim holds, the paper is a solid contribution: it defines a more realistic camera-motion task, releases a new benchmark, and shows that geometric knowledge from a 3D foundation model can be compressed into lightweight per-frame tokens and discarded at inference. The strengths are the controlled comparison (same backbone, same training data, only the teacher's presence at inference varied), the lambda_cam=0 partial control in Figure 4, the external transfer evaluation, and the careful latency/memory measurements. The main weakness is that benchmark reliability is not quantified and all experiments are single runs, leaving the headline 'matches' claim under-supported.

major comments (3)
  1. [Section 3.2 / Appendix C-D] The benchmark's ground truth is load-bearing for every reported model ranking, including the central CamDistill-equals-CamInject comparison, yet no inter-annotator agreement statistic is provided. The protocol describes five annotators, cross-review, and discarding unresolved clips, but report no Cohen's kappa or Krippendorff's alpha for either segment boundaries or the 20-label sets, nor a reliability analysis for the hard distinctions (Dolly versus Zoom, Pan versus Truck) that the taxonomy explicitly requires. Additionally, Appendix C says annotators 'independently replace the pseudo labels,' which suggests they saw model-generated suggestions; this could introduce anchoring bias. Please report agreement statistics on a per-family or per-pair basis, clarify whether pseudo-labels were visible during annotation, and discuss the selection effect of discarding unresolved clips.
  2. [Section 5.2, Table 2] All results are single runs, and the claim that CamDistill 'matches' CamInject is based on differences of 0.0 to 0.5 micro F1 points (67.5 vs 67.5 for 4B; 67.8 vs 68.3 for 8B) and 0.4 to 0.5 SegDet@0.5 points. Without multiple seeds, confidence intervals, or a paired significance test, these differences are within typical run-to-run noise, so the equality could be coincidental. Please provide variance estimates or significance tests for the key CamDistill-versus-CamInject comparisons on the 4B and 8B backbones.
  3. [Abstract / Section 5.2] The unqualified abstract statement that CamDistill 'matches the accuracy of direct feature injection' overstates the 8B results: on the 8B backbone CamDistill trails CamInject by 0.5 micro F1 and 1.3 macro F1 (Table 2). The body text acknowledges this, but the abstract and conclusion should either qualify the claim (e.g., 'matches on the 4B backbone with a small gap at 8B') or provide statistical evidence that the 8B gap is within noise.
minor comments (5)
  1. [Section 5.2, Table 2] The Cam-Motion-7B row states that the model yields valid outputs for only 19 of 4,229 clips; please clarify whether the reported scores are computed on those 19 clips alone and specify which subset, as the current wording is ambiguous.
  2. [Figure 4] The x-axis includes a tick at lambda_cam = 0 but this value is not labeled; please label it explicitly and report the lambda=0 result in the text or a table, since it is the key control for isolating the distillation objective.
  3. [Section 4.2, Eqs. (2)-(3)] The notation c_i^{(m)}, x_i^{(ell_m)}, and the block index m should be defined in the text before their first use in the equations to avoid ambiguity.
  4. [Appendix K, Algorithm 1] Line 7, 'Save the post-frame-wise cross-attention tokens in the last block as the frame-level branch,' is ambiguous because the loop notation suggests it executes every iteration; reword to state that the save occurs after the final block.
  5. [General] The manuscript inconsistently uses both CAMCHOREO and CamChoreo (and similarly CAMDISTILL/CamDistill); please unify the capitalization style throughout.

Circularity Check

0 steps flagged · score 0.0 of 10

No meaningful circularity: CamDistill's teacher is an externally pretrained 3D model, the benchmark labels are human annotations on independently collected clips, and the headline match is an empirical result, not a fitted identity.

full rationale

The paper's central claim is empirical: CamDistill (distilled camera tokens) matches CamInject (direct teacher-feature injection) on a held-out benchmark. Nothing in Eq. (4) or in the GCTE architecture forces this equality; the student is trained with an auxiliary cosine loss against cached VGGT-Omega tokens, while CamInject uses the actual teacher tokens at inference, and the 67.5-vs-67.5 result is a measured comparison reported in Table 3. The benchmark CamChoreo is not defined in terms of the proposed model: Section 3.2 states that five annotators label every clip from scratch and that clips with unresolved disagreements are discarded, and Section 5.1 and Appendix G state that the training clips come from a disjoint Tencent Video pool annotated with the same protocol, while evaluation uses the separately collected YouTube clips. The distillation target is an external frozen geometry model, not a quantity derived from the benchmark labels or from CamDistill's own outputs. External CameraBench and CMVQA results in Table 4 provide independent checks of transfer. The two reviewer-flagged issues--absence of inter-annotator agreement statistics and validation-set tuning of the distillation weight--are data-quality or statistical concerns, not circularity: a biased annotation protocol would affect all models similarly and does not make the CamDistill-versus-CamInject comparison true by construction. There are no load-bearing self-citations, no imported uniqueness theorems, and no fitted parameter renamed as a prediction.

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

The central method depends on a small number of tuned hyperparameters, the reliability of expert annotations, and the validity of VGGT-Omega camera tokens as a distillation target. No new physical entities are introduced.

free parameters (5)
  • distillation weight lambda_cam = 0.05
    Peak of the validation sensitivity sweep in Figure 4; this weight controls how much the student camera tokens imitate the VGGT-Omega teacher, and the paper's efficiency claim depends on the distillation being useful at a nonzero value.
  • GCTE depth and tapped ViT layers = 4 blocks, layers {1,5,9,13} for 4B; 8 blocks, layers {1,3,5,7,9,11,13,15} for 8B
    Selected by ablations in Table 5 and Appendix G, not derived from theory; the method's performance is sensitive to this choice.
  • camera token position = before each frame's visual tokens
    Ablation choice in Table 5; placing tokens before visual tokens adds 2.8 micro F1 over the alternative.
  • teacher model = VGGT-Omega
    Ablation in Table 5; replacing VGGT with VGGT-Omega adds 1.7 micro F1.
  • video sampling rate and max frames = 5 FPS, up to 100 frames
    Evaluation and training protocol (Appendix G); temporal grounding resolution depends on this choice.
assumptions (5)
  • domain assumption VGGT-Omega camera tokens encode camera geometry in a way that is useful for motion-label grounding and can be imitated from frozen intermediate ViT features.
    The entire CamDistill mechanism in Sections 4.1-4.3 assumes this; the paper validates it empirically but provides no independent analysis of teacher token quality.
  • domain assumption Expert annotators can reliably segment motion-consistent intervals and assign direction-aware labels at 0.1-second resolution.
    Appendix D describes the protocol but reports no inter-annotator agreement statistics, so label reliability is assumed rather than demonstrated.
  • domain assumption The same 20-label taxonomy and annotation protocol transfer between the Tencent Video training set and the YouTube benchmark.
    Appendix G states the two sets are disjoint and use the same protocol, but cross-source label consistency is not measured.
  • standard math Standard transformer attention, cosine-distance distillation, and next-token SFT are valid building blocks for this task.
    Used throughout Sections 4.2-4.3 without new mathematical derivation; these are background ML operations.
  • domain assumption Frame-level and segment-level metrics with Hungarian matching capture task quality.
    Appendix E defines the metrics; the choice of exact label-set matching for SegDet is a strict and domain-specific decision.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Temporally Grounded Compositional Camera Motion Understanding via Geometric Knowledge Distillation." pith.science (2026). https://pith.science/paper/P7J2XQ2M

@misc{pith2026260810932,
  author       = {Pith},
  title        = {Pith review of: Temporally Grounded Compositional Camera Motion Understanding via Geometric Knowledge Distillation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/P7J2XQ2M}},
  note         = {Machine review of arXiv:2608.10932}
}
read the original abstract

Understanding camera motion is fundamental to video perception, with applications in spatial intelligence and controllable video generation. Multimodal large language models (MLLMs) provide a natural interface for this task, but existing work typically assigns one or more labels to an entire clip. Such clip-level recognition overlooks two defining properties of real camera motion: it can change within a shot, and multiple movements can occur simultaneously. We therefore formulate camera-motion understanding as temporally grounded, compositional recognition, which requires a model to localize motion-consistent intervals and identify every movement active within each interval. We introduce CamChoreo, a benchmark of 4,229 real single-shot clips with expert-annotated temporal segments. Its annotations use a compact vocabulary of 20 direction-aware labels, and nearly half of the segments contain compound camera motion, with multiple movement primitives active simultaneously. Recognizing such fine-grained, compositional motion is hard for current MLLMs, whose visual encoders emphasize semantic content rather than the geometric evidence on which camera motion depends. Directly injecting features from a frozen 3D foundation model addresses this gap, but requires running the expensive geometry model on every input; we refer to this baseline as CamInject. We instead propose CamDistill, which distills the same geometric knowledge into lightweight camera tokens during training and removes the 3D model at inference. CamDistill matches the accuracy of direct feature injection without running the 3D teacher at inference. Together, CamChoreo and CamDistill advance camera-motion understanding from clip-level labeling to temporally grounded, compositional recognition. Project page: https://ddz16.github.io/cammotion.github.io/.

Figures

Figures reproduced from arXiv: 2608.10932 by the authors.

Figure 2
Figure 2. CAMCHOREO statistics. (A) video duration, (B) number of segments per clip, (C) number of simultaneous movements per segment, (D) the long-tailed distribution of the 20 direction-aware labels, and (E) the nine content domains. 3.2 DATA CURATION Collection and filtering. We collect YouTube footage from nine content domains and split each video into single shots at hard cuts with TransNetV2 (Soucek & Lokoc, 2024). The … view at source ↗
Figure 3
Figure 3. CAMINJECT versus CAMDISTILL. (a) CAMINJECT runs a frozen 3D foundation model (VGGT-Ω) beside the vision encoder and inserts each frame’s teacher camera token just before that frame’s visual tokens in the LLM input, so the 3D model is required at inference. (b) CAMDISTILL instead trains a lightweight GCTE, a stack of M alternating frame-wise cross-attention and global camera self-attention blocks, to predict the stud… view at source ↗
Figure 4
Figure 4. Distillation-weight sensitivity. CAMDISTILL-4B peaks at λcam = 0.05 [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figures from the paper (14 more)
Figure 5
Figure 5. Figure 5: summarizes the complete curation pipeline, from source discovery and automatic filtering to class balancing and human quality control [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Training loss curves. (a) Next-token SFT loss and (b) camera-token distillation loss for CAMDIS￾TILL-4B (solid) and CAMDISTILL-8B (dashed) over normalized training progress. Both backbones converge to similar final values (SFT ≈ 0.003, distillation ≈ 0.04), indicating …
Figure 7
Figure 7. Figure 7: Camera-motion co-occurrence in CAMCHOREO. Left: direction-aware co-occurrence matrix. Right: most frequent label pairs. Static Unstable Pan-L Pan-R Tilt-U Tilt-D Roll-CW Roll-CCW Truck-L Truck-R Pedestal-U Pedestal-D Dolly-In Dolly-Out Zoom-In Zoom-Out Focus Follow Arc…
Figure 8
Figure 8. Figure 8: Error analysis of CAMDISTILL-8B. Left: row-normalized frame-level confusions. Right: segment￾level error decomposition after IoU ≥ 0.5 matching. is removed, the resulting inference cost is close to that of the SFT backbone, consistent with the latency and memory measur…
Figure 9
Figure 9. Figure 9: Per-class frame-level F1 of CAMDISTILL-8B, sorted by score and colored by motion family. 1 2 3+ # simultaneous camera motions in frame 0 20 40 60 80 100 Frame-level F1 (%) 72.1 63.9 65.8 (a) F1 vs. compositional density 1 2 3 4 5+ # ground-truth segments per video 0 20…
Figure 10
Figure 10. Figure 10: Performance versus task complexity. (a) Frame-level F1 decreases as more camera motions co￾occur; (b) mean per-video F1 decreases as the clip is split into more temporal segments. Per-class difficulty and the long tail [PITH_FULL_IMAGE:figures/full_fig_p024_10.png]
Figure 11
Figure 11. Figure 11: Boundary precision. Signed offset between matched predicted and ground-truth segment boundaries [PITH_FULL_IMAGE:figures/full_fig_p025_11.png]
Figure 12
Figure 12. Figure 12: Qualitative predictions of CAMDISTILL-8B on CAMCHOREO (part 1). Matched GT and Pred segments (by temporal IoU) share a color, unmatched segments are gray, and each segment lists its movements. 26 [PITH_FULL_IMAGE:figures/full_fig_p026_12.png]
Figure 13
Figure 13. Figure 13: Qualitative predictions on CAMCHOREO (part 2). Conventions as in [PITH_FULL_IMAGE:figures/full_fig_p027_13.png]
Figure 14
Figure 14. Figure 14: Qualitative predictions on CAMCHOREO (part 3). Conventions as in [PITH_FULL_IMAGE:figures/full_fig_p028_14.png]
Figure 15
Figure 15. Figure 15: Qualitative predictions on CAMCHOREO (part 4). Conventions as in [PITH_FULL_IMAGE:figures/full_fig_p029_15.png]
Figure 16
Figure 16. Figure 16: Qualitative predictions on CAMCHOREO (part 5). Conventions as in [PITH_FULL_IMAGE:figures/full_fig_p030_16.png]
Figure 17
Figure 17. Figure 17: Qualitative predictions on CAMCHOREO (part 6). Conventions as in [PITH_FULL_IMAGE:figures/full_fig_p031_17.png]
Figure 18
Figure 18. Figure 18: Failure cases of CAMDISTILL-8B. Top to bottom: a lateral Truck predicted as Pan; a forward Dolly predicted as Zoom; a hand-held Follow predicted as Static or Pan; and Unstable footage predicted as Roll and over-segmented into short intervals. Conventions as in [PITH_…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 5 canonical work pages

  1. [1]

    Layer normalization.arXiv preprint arXiv:1607.06450,

    Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. Layer normalization.arXiv preprint arXiv:1607.06450,

    arXiv

  2. [7]

    Llava-onevision: Easy visual task transfer.arXiv preprint arXiv:2408.03326, 2024a

    Bo Li, Yuanhan Zhang, Dong Guo, Renrui Zhang, Feng Li, Hao Zhang, Kaichen Zhang, Peiyuan Zhang, Yanwei Li, Ziwei Liu, et al. Llava-onevision: Easy visual task transfer.arXiv preprint arXiv:2408.03326, 2024a. Moyang Li, Zihan Zhu, Marc Pollefeys, and Daniel Barath. Droid-slam in the wild.arXiv preprint arXiv:2603.19076,

    arXiv

  3. [8]

    Can video generation replace cinematographers? Research on the cinematic language of generated video

    Xiaozhe Li, Kai Wu, Siyi Yang, YiZhan Qu, Guohua Zhang, Zhiyu Chen, Jiayao Li, Jiangchuan Mu, Xiaobin Hu, Wen Fang, et al. Can video generation replace cinematographers? research on the cinematic language of generated video.arXiv preprint arXiv:2412.12223, 2024b. Zhiqiu Lin, Siyuan Cen, Daniel Jiang, Jay Karhade, Hewei Wang, Chancharik Mitra, Yu Tong Tiff...

    work page Pith review arXiv

  4. [9]

    Openai gpt-5 system card.arXiv preprint arXiv:2601.03267,

    Aaditya Singh, Adam Fry, Adam Perelman, Adam Tart, Adi Ganesh, Ahmed El-Kishky, Aidan McLaughlin, Aiden Low, AJ Ostrow, Akhila Ananthram, et al. Openai gpt-5 system card.arXiv preprint arXiv:2601.03267,

    arXiv

  5. [10]

    Per-class difficulty and the long tail.Figure 9 reports frame-level F1 for all 20 direction-aware labels

    Performance versus task complexity.(a) Frame-level F1 decreases as more camera motions co- occur; (b) mean per-video F1 decreases as the clip is split into more temporal segments. Per-class difficulty and the long tail.Figure 9 reports frame-level F1 for all 20 direction-aware labels. Performance is uneven and broadly follows class frequency: the more fre...

    work page 2000

  6. [13]

    Cambrian-P: Pose-Grounded Video Understanding

    Jihan Yang, Zifan Zhao, Xichen Pan, Shusheng Yang, Junyi Zhang, Bingyi Kang, Hu Xu, and Saining Xie. Cambrian-p: Pose-grounded video understanding.arXiv preprint arXiv:2605.22819,

    work page Pith review arXiv

  7. [15]

    Llava-video: Video instruction tuning with synthetic data.arXiv preprint arXiv:2410.02713,

    Yuanhan Zhang, Jinming Wu, Wei Li, Bo Li, Zejun Ma, Ziwei Liu, and Chunyuan Li. Llava-video: Video instruction tuning with synthetic data.arXiv preprint arXiv:2410.02713,

    arXiv

  8. [16]

    These pipelines can be slow and brittle under low texture or pure rotation

    and DROID-SLAM (Teed & Deng, 2021), recover trajectories through feature matching and geometric optimization. These pipelines can be slow and brittle under low texture or pure rotation. They are also sensitive to dynamic subjects, which are common in film, television, and vlog footage: when a moving person occupies much of the frame, feature matching may ...

    work page 2021

Show all 18 references
  1. [17]

    Large annotated resources such as SpatialVID (Wang et al., 2025a) further support progress in video geometry

    improves SLAM robustness in dynamic environments. Large annotated resources such as SpatialVID (Wang et al., 2025a) further support progress in video geometry. We selectVGGT / VGGT-Ωas teachers for three reasons. First, their feed-forward inference avoids per-scene optimizatio...

  2. [2014]

    Geometry-guided camera motion under- standing in videollms.arXiv preprint arXiv:2603.13119,

    Haoan Feng, Sri Harsha Musunuri, and Guan-Ming Su. Geometry-guided camera motion under- standing in videollms.arXiv preprint arXiv:2603.13119,

  3. [2015]

    Vipe: Video pose engine for 3d geometric perception.arXiv preprint arXiv:2508.10934,

    Jiahui Huang, Qunjie Zhou, Hesam Rabeti, Aleksandr Korovko, Huan Ling, Xuanchi Ren, Tian- chang Shen, Jun Gao, Dmitry Slepichev, Chen-Hsuan Lin, et al. Vipe: Video pose engine for 3d geometric perception.arXiv preprint arXiv:2508.10934,

    arXiv

  4. [2016]

    Recammaster: Camera-controlled generative rendering from a single video

    Jianhong Bai, Menghan Xia, Xiao Fu, Xintao Wang, Lianrui Mu, Jinwen Cao, Zuozhu Liu, Haoji Hu, Xiang Bai, Pengfei Wan, et al. Recammaster: Camera-controlled generative rendering from a single video. In2025 IEEE/CVF International Conference on Computer Vision (ICCV), pp. 14834–...

    arXiv

  5. [2017]

    Spatialvid: A large-scale video dataset with spatial annotations.arXiv preprint arXiv:2509.09676, 2025a

    Jiahao Wang, Yufeng Yuan, Rujie Zheng, Youtian Lin, Jian Gao, Lin-Zhuo Chen, Yajie Bao, Yi Zhang, Chang Zeng, Yanxi Zhou, et al. Spatialvid: A large-scale video dataset with spatial annotations.arXiv preprint arXiv:2509.09676, 2025a. Jianyuan Wang, Minghao Chen, Nikita Karaev,...

  6. [2021]

    Siglip 2: Multilingual vision-language encoders with improved semantic understanding, localization, and dense features.arXiv preprint arXiv:2502.14786,

    Michael Tschannen, Alexey Gritsenko, Xiao Wang, Muhammad Ferjad Naeem, Ibrahim Alabdul- mohsin, Nikhil Parthasarathy, Talfan Evans, Lucas Beyer, Ye Xia, Basil Mustafa, et al. Siglip 2: Multilingual vision-language encoders with improved semantic understanding, localization, an...

    arXiv

  7. [2023]

    On the generalization capacities of mllms for spatial intelligence.arXiv preprint arXiv:2603.06704,

    Gongjie Zhang, Wenhao Li, Quanhao Qian, Jiuniu Wang, Deli Zhao, Shijian Lu, and Ran Xu. On the generalization capacities of mllms for spatial intelligence.arXiv preprint arXiv:2603.06704,

  8. [2024]

    Internvl3.5: Advancing open-source multimodal models in versatility, reasoning, and efficiency.arXiv preprint arXiv:2508.18265, 2025c

    Weiyun Wang, Zhangwei Gao, Lixin Gu, Hengjun Pu, Long Cui, Xingguang Wei, Zhaoyang Liu, Linglin Jing, Shenglong Ye, Jie Shao, et al. Internvl3.5: Advancing open-source multimodal models in versatility, reasoning, and efficiency.arXiv preprint arXiv:2508.18265, 2025c. Xinran Wa...

    2026 arXiv

  9. [2025]

    An image is worth 16x16 words: Transformers for image recognition at scale.arXiv preprint arXiv:2010.11929,

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale.arXiv preprint arXiv:...

    2010 arXiv

  10. [2026]

    Distilling the knowledge in a neural network.arXiv preprint arXiv:1503.02531,

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the knowledge in a neural network.arXiv preprint arXiv:1503.02531,

    arXiv

Pith tools

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