Pith. sign in

REVIEW 4 major objections 5 minor 40 references

A Controlled Study of Self-Supervised Image and Video Pretraining under Limited Resources

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Under a tight 80,000-update budget on Kinetics-700 with a shared ViT-B/16 backbone, DINOv2-style pretraining is the strongest all-round self-supervised recipe, and adding VideoMAE improves semantics while sacrificing some video geometry.

desk verdict A genuinely controlled SSL benchmark worth refereeing, but the matched-compute claim only matches optimizer updates, not FLOPs, and the headline gains are small point estimates without variance. read the letter →

arxiv 2608.13183 v1 pith:VAV7PNA7 submitted 2026-08-13 cs.CV

Brunó B. Englert , Gijs Dubbelman This is my paper · ORCID
classification cs.CV
keywords self-supervisedlearningvisualfoundationmodelsresource-limitedpretrainingDINOv2VideoMAEimage-videojointtrainingsemantic-geometrictradeoffcontrolledbenchmarking
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 asks which self-supervised learning objective yields the most useful visual backbone when pretraining compute is scarce. The authors run a controlled comparison in which every method shares the same source data (Kinetics-700), the same ViT-B/16 image encoder, and the same 80,000 optimizer updates, varying only the SSL objective. They find that DINOv2-style teacher–student pretraining is the strongest all-round performer on image classification, segmentation, and depth, while VideoMAE-style masked reconstruction wins on video motion and geometry tasks such as point tracking and relative camera pose. Combining the two objectives produces a single frozen backbone that beats both specialists on semantics and beats plain DINOv2 on pose, but falls short of VideoMAE on geometry—a tradeoff the paper quantifies. The practical point is that under limited resources, the choice and combination of SSL objectives matter more than any single benchmark ranking.

What carries the argument

The load-bearing mechanism is the controlled pretraining framework: a shared ViT-B/16 image encoder that is applied either to single frames or framewise to 8-frame clips, an optional temporal neck that performs full bidirectional space–time self-attention over all patch and CLS tokens, and a fixed schedule of 80,000 optimizer updates on Kinetics-700. Image SSL objectives supervise the shared encoder directly; video SSL objectives pass framewise tokens through the temporal neck and backpropagate into both encoder and neck; joint runs simply sum the losses. The DINOv2 objective—teacher–student cross-entropy over Sinkhorn-Knopp prototype assignments plus iBOT-style masked patch prediction with an EMA teacher—is the semantic anchor, and its combination with VideoMAE's masked spatiotemporal token reconstruction is the main joint recipe. The framework isolates the SSL objective as the only changing variable, and a neck-reuse experiment (evaluating video tasks with and without the frozen pretrained neck) localizes geometric-temporal structure in the neck while semantics live in the image encoder.

What would settle it

Repeat the same 80k-update, Kinetics-700, ViT-B/16 comparison but match total FLOPs or wall-clock time across objectives instead of update count; if DINOv2 no longer ranks first on the image tasks or VideoMAE no longer ranks first on pose and tracking, the reported ranking is an artifact of the update-count budget. A second check: run the DINOv2 + VideoMAE joint recipe under a wall-clock budget equal to that of standalone VideoMAE (allowing the joint recipe fewer updates) and see whether its geometric deficit persists or widens.

Watch

Extended reading notes

Core claim

Under a matched, limited pretraining budget—80k optimizer updates on the Kinetics-700 dataset with a ViT-B/16 image encoder—DINOv2-style pretraining consistently provides the strongest overall performance, ranking first on all six image evaluations (ImageNet-1K classification, Pascal VOC, Cityscapes, ADE20K segmentation, NYUv2 and KITTI depth) and on appearance-driven video classification on Kinetics-700. VideoMAE-style reconstruction, by contrast, ranks first on the geometric and temporal video tasks: SSV2 recognition, relative camera pose on RE10K, and point tracking on MOVi-F. Jointly training DINOv2 with VideoMAE improves all six classification and segmentation evaluations beyond both standalone methods, improves pose relative to DINOv2 alone, but degrades tracking and pose relative to VideoMAE alone, revealing a semantic-versus-geometric tradeoff. The authors read these results as evidence that image and video SSL objectives are complementary under resource constraints, and that careful objective selection and combination can produce strong general-purpose representations within a small budget.

Load-bearing premise

The controlled comparison treats 80,000 matched optimizer updates as a fair stand-in for 'same compute budget', even though the paper notes this does not equate FLOPs or wall-clock time across objectives.

Editorial extensions

If this is right

  • Under an 80k-update budget, practitioners should expect DINOv2-style pretraining to provide the best all-round semantic backbone and VideoMAE-style reconstruction the best geometry-focused backbone.
  • Joint DINOv2 + VideoMAE training yields one frozen representation that outperforms both specialists on semantic tasks and outperforms DINOv2 alone on camera-pose estimation, making it a strong default for mixed image-and-video use.
  • The temporal neck trained by video objectives is the main repository of geometric-temporal structure: removing it hurts video geometry tasks much more than video semantic tasks.
  • Objective compatibility matters under short budgets: adding masked reconstruction (I-MAE) to DINOv2 modestly improves segmentation and classification, while adding a diffusion denoising loss (I-Diffusion) mostly hurts transfer.

Reading between the lines

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

  • If the temporal neck is the main repository of geometric structure, then co-training that neck with a geometry-specific loss—such as a point-tracking or relative-pose objective—could recover VideoMAE-level pose accuracy while keeping DINOv2 semantics, a combination the paper does not test.
  • The update-count budget likely favors objectives that learn quickly per update (feature-space methods) over those that need long schedules (diffusion), so the ranking may not extrapolate to larger budgets; a FLOP-matched repeat could reorder the lower half of the table.
  • The authors never sweep the relative weight of the DINOv2 and VideoMAE losses in the joint recipe, so the single reported joint point may understate what the combination could achieve with a different loss balance.
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.

Referee Report

4 major / 5 minor

Summary. The paper reports a controlled empirical comparison of seven self-supervised pretraining objectives (DINOv2, I-MAE, V-MAE, I-JEPA, V-JEPA, I-Diffusion, V-Diffusion) plus DINOv2-anchored joint variants, all trained on Kinetics-700 with a ViT-B/16 backbone under an 80k-optimizer-update budget. The authors evaluate frozen representations on image classification, semantic segmentation, monocular depth, action recognition, point tracking, and relative camera pose estimation. The central claims are that DINOv2 gives the strongest semantic representation under this limited budget, V-MAE gives the strongest geometric/temporal representation, and joint DINOv2+V-MAE training improves semantic image and video tasks while degrading tracking and camera-pose performance, revealing a semantic/geometric tradeoff.

Significance. If the results hold, the paper provides a useful benchmark for resource-limited visual foundation model development, with a broader task suite than most SSL comparisons and a genuinely unified architecture for image and video objectives. The strengths are the controlled data and backbone, the inclusion of four objective families, the joint-training experiments, and the public code release. However, the claims currently rest on update-count matching rather than true compute matching, on single-run point estimates with no variance analysis, and on video-task comparisons that give video methods access to additional pretrained temporal parameters. These issues are load-bearing for the paper's central ranking and tradeoff conclusions, so the present version needs substantive revision.

major comments (4)
  1. [Sec. 4.1, Table 2] The title and abstract promise comparison under 'matched ... compute budgets,' but the actual budget is 80k optimizer updates, and Sec. 4.1 states this 'does not equate FLOPs or wall-clock time.' Because video objectives process eight-frame clips through the shared ViT plus a temporal neck, their per-update FLOPs are substantially higher than image-only objectives; Table 2's reported range of 144-224 GPU-hours per run confirms that total resource use varies across methods. As a result, the ranking of objectives, and in particular the conclusion that V-MAE is strongest on geometric tasks, may reflect unequal compute rather than objective quality. Please either match total FLOPs or wall-clock time, or explicitly reframe the paper's claims as holding under a matched-update budget and report per-method FLOPs/GPU-hours so readers can assess the sensitivity.
  2. [Tables 3-7] All results are single-run point estimates with no variance, error bars, or significance tests. Several of the central quantitative claims rest on small differences: Table 5 reports +0.6 ImageNet top-1, +0.4 to +0.7 mIoU, and -0.02/+0.06 RMSE for DINOv2+I-MAE, and Table 6 reports +0.5 ImageNet top-1 and -0.3 MOVi-F Jaccard for DINOv2+V-MAE relative to DINOv2. These differences are within the range where training noise could alter the ranking. Please report multiple seeds with standard deviations, and use a significance test or at least a clear threshold for what counts as an improvement/degradation.
  3. [Sec. 4.2, Table 4; Sec. 4.4, Table 6] The video-task comparisons are not matched in pretrained parameters: video SSL methods are evaluated with their frozen pretrained temporal neck, while image-only methods (DINOv2, I-MAE, etc.) have no pretrained neck and must learn a lightweight neck from task labels. Table 7 shows that the pretrained neck contributes substantially to V-MAE's video performance (e.g., K700 43.3 vs. 52.8, RE10K rotation 84.4 vs. 90.7). Thus the conclusion that V-MAE is the strongest geometric objective conflates the effect of the SSL objective with the effect of having extra pretrained temporal parameters. Please report the encoder-only variant consistently for all methods in the main video comparisons, and separate the effect of the temporal neck in the claims.
  4. [Abstract; Table 6] The abstract states that combining DINOv2 with VideoMAE 'substantially improves image classification and segmentation performance, but degrades video tracking and camera-pose estimation performance.' Relative to DINOv2, the joint recipe improves ImageNet top-1 by only +0.5 and mIoU by +0.3 to +1.4, and tracking decreases by 0.3 Jaccard points; these are modest and may not be significant given the lack of variance estimates. The pose changes are large (+36.4 rotation accuracy), but the overall tradeoff claim should be reworded to match the magnitude of the measured differences and should be supported by statistical evidence.
minor comments (5)
  1. [Table 2] The formatting '224 2', '2×162', and similar entries appears to be missing superscripts; please render dimensions consistently (e.g., 224², 16×224²).
  2. [Sec. 4.5] The sentence 'All evaluation experiments use a lightweight temporal neck whose weights are not pretrained and are learned solely from the labels of the evaluation datasets' appears to contradict the pretrained-neck variant (✓) in Table 7; please clarify that this statement applies to the encoder-only variant only.
  3. [Table 7] For V-MAE + DINOv2, the MOVi-F Jaccard is 66.9 without the neck and 65.7 with it, which is a counterintuitive direction; please comment on whether this is within expected noise and consider reporting more decimal places.
  4. [Sec. 3.4] The downstream evaluation protocol does not specify the lightweight head architectures, optimization hyperparameters, or training budgets for each task; please add these details in an appendix or supplementary material for reproducibility.
  5. [Table 5] Some entries appear rounded to the same value (e.g., ADE20K 25.0 vs. 25.0; NYUv2 0.97 vs. 0.95), making the reported changes less informative; please report additional significant digits or include confidence intervals.

Circularity Check

0 steps flagged · score 0.0 of 10

Empirical controlled comparison with no circular derivation; all headline results are out-of-sample probes of frozen representations.

full rationale

This is an empirical benchmark study rather than a derivation. Each SSL objective is defined in Sec. 3.1 from standard external losses (DINOv2 cross-entropy over Sinkhorn assignments, JEPA L1 in representation space, MAE pixel MSE, and diffusion denoising), and downstream evaluations in Sec. 4 freeze the pretrained modules and train only lightweight heads on held-out task labels. The central claims—DINOv2 ranking first in Table 3, V-MAE leading geometry in Table 4, and DINOv2+V-MAE improving semantic tasks while degrading pose/tracking in Table 6—are measured outcomes, not restatements of the loss definitions, fitted constants, or self-citation chains. The only self-citation, reference [24] (VidEoMT, co-authored by Dubbelman), supports a background observation about video systems starting from strong image encoders and is not load-bearing for any result. The admitted mismatch between optimizer-update budget and FLOPs/wall-clock time (Sec. 4.1) is a resource-matching validity concern, not a circularity: it does not make any reported quantity equivalent to an input by construction. No circular step can be exhibited, so the score is 0.

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

The paper's claims are empirical, so the ledger holds no fitted constants or invented entities. The load-bearing inputs are protocol choices: the 80k-update budget, K700 as source data, ViT-B/16 plus temporal neck as architecture, and the frozen-backbone evaluation scheme. These are domain assumptions rather than mathematical axioms or postulated entities.

assumptions (5)
  • domain assumption Matched optimizer-update count measures matched compute budget for all SSL objectives.
    Sec 4.1 states that the budget is fixed by optimizer updates but that FLOPs and wall-clock time are not equated; all conclusions rest on updates being the relevant resource.
  • domain assumption K700 is a fair common source dataset for image and video SSL objectives.
    Sec 3.3 trains every method on K700; this assumes one video dataset can support both frame-level and clip-level objectives without bias.
  • domain assumption A ViT-B/16 encoder with a full space-time attention neck is a sufficient common architecture to reveal objective differences.
    Sec 3.2 fixes this architecture across methods; rankings may differ with other capacities or temporal designs.
  • domain assumption Frozen encoder plus lightweight heads is a valid probe of representation quality.
    Sec 3.4 freezes pretrained modules for all downstream tasks; the conclusions describe this protocol, not full fine-tuning.
  • domain assumption The 80k-update schedule is representative of the limited-resource regime the paper targets.
    Sec 3.3 and Table 2 set this schedule; the paper explicitly limits claims to this budget.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Controlled Study of Self-Supervised Image and Video Pretraining under Limited Resources." pith.science (2026). https://pith.science/paper/VAV7PNA7

@misc{pith2026260813183,
  author       = {Pith},
  title        = {Pith review of: A Controlled Study of Self-Supervised Image and Video Pretraining under Limited Resources},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VAV7PNA7}},
  note         = {Machine review of arXiv:2608.13183}
}
read the original abstract

Visual foundation models are a cornerstone of image and video understanding but typically require large amounts of data and computation. The current scale required for pretraining visual foundation models may be unsustainable or unnecessary, and significant benefits arise when effective models can be obtained with fewer resources. To better understand how self-supervised learning (SSL) objectives behave under resource constraints, we conduct a controlled study of image and video SSL objectives under matched data, architecture, and compute budgets. We compare contrastive, reconstruction, feature-prediction, and diffusion objectives and evaluate both standalone and jointly trained image-video SSL formulations across a diverse set of image and video understanding tasks. Our results show that DINOv2-style pretraining consistently provides the strongest overall performance under limited resources. Furthermore, combining DINOv2 with video SSL objectives such as VideoMAE substantially improves image classification and segmentation performance, but degrades video tracking and camera-pose estimation performance, revealing an important tradeoff between semantic and geometric representation learning. These findings suggest that combining image and video SSL objectives can be beneficial in resource-limited settings, while highlighting the need for improved methods that better balance semantic, temporal, and geometric supervision.

Figures

Figures reproduced from arXiv: 2608.13183 by the authors.

Figure 1
Figure 1. Method architecture overview. The image-only encoder is the common representation used for all image tasks, which matches image-task compute and pa￾rameter count across image and video SSL. Video objectives add a temporal neck after the framewise encoder; this neck can be reused or omitted during video downstream evaluation. number of patch tokens per frame, and C is the encoder width. The temporal neck Nϕ preserves… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 30 canonical work pages

  1. [1]

    arXiv preprint arXiv:2502.07811 (2025),https://arxiv.org/abs/ 2502.07811

    Ahamed, S.A., Gunawardhana, M., David, L., Sidorov, M., Harari, D., Khan, M.H.: CrossVideoMAE: Self-supervised image-video representation learning with masked autoencoders. arXiv preprint arXiv:2502.07811 (2025),https://arxiv.org/abs/ 2502.07811

    arXiv 2025

  2. [2]

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

    Assran, M., Duval, Q., Misra, I., Bojanowski, P., Vincent, P., Rabbat, M., LeCun, Y., Ballas, N.: Self-supervised learning from images with a joint-embedding pre- dictive architecture. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 15619–15629 (2023)

    work page 2023

  3. [3]

    In: Proceed- ings of the 39th International Conference on Machine Learning (ICML)

    Baevski, A., Hsu, W.N., Xu, Q., Babu, A., Gu, J., Auli, M.: data2vec: A general framework for self-supervised learning in speech, vision and language. In: Proceed- ings of the 39th International Conference on Machine Learning (ICML). Proceed- ings of Machine Learning Research, vol. 162, pp. 1298–1312. PMLR (2022)

    work page 2022

  4. [4]

    In: International Conference on Learning Representations (ICLR) (2022), https://openreview.net/forum?id=p-BhZSz59o4

    Bao, H., Dong, L., Piao, S., Wei, F.: BEiT: BERT pre-training of image trans- formers. In: International Conference on Learning Representations (ICLR) (2022), https://openreview.net/forum?id=p-BhZSz59o4

    work page 2022

  5. [5]

    Transactions on Machine Learning Research (2024),https://openreview

    Bardes, A., Garrido, Q., Ponce, J., Chen, X., Rabbat, M., LeCun, Y., Assran, M., Ballas, N.: Revisiting feature prediction for learning visual representations from video. Transactions on Machine Learning Research (2024),https://openreview. net/forum?id=QaCCuDfBk2

    work page 2024

  6. [6]

    In: International Conference on Learning Rep- resentations (ICLR) (2022),https://openreview.net/forum?id=xm6YD62D1Ub

    Bardes, A., Ponce, J., LeCun, Y.: VICReg: Variance-invariance-covariance regular- ization for self-supervised learning. In: International Conference on Learning Rep- resentations (ICLR) (2022),https://openreview.net/forum?id=xm6YD62D1Ub

    work page 2022

  7. [7]

    In: Advances in Neural Information Processing Systems (NeurIPS)

    Caron, M., Misra, I., Mairal, J., Goyal, P., Bojanowski, P., Joulin, A.: Unsuper- vised learning of visual features by contrasting cluster assignments. In: Advances in Neural Information Processing Systems (NeurIPS). vol. 33 (2020) 16 B.B. Englert and G. Dubbelman

    work page 2020

  8. [8]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Caron, M., Touvron, H., Misra, I., Jégou, H., Mairal, J., Bojanowski, P., Joulin, A.: Emerging properties in self-supervised vision transformers. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 9650–9660 (2021)

    2021

Show all 40 references
  1. [9]

    arXiv preprint arXiv:2412.15212 (2024)

    Carreira, J., Gokay, D., King, M., Zhang, C., Rocco, I., Mahendran, A., Keck, T.A., Heyward, J., Koppula, S., Pot, E., Erdogan, G., Hasson, Y., Yang, Y., Greff, K., Le Moing, G., van Steenkiste, S., Zoran, D., Hudson, D.A., Vélez, P., Polanía, L., Friedman, L., Duvarney, C., e...

    2024 arXiv

  2. [10]

    arXiv preprint arXiv:1907.06987 (2019)

    Carreira, J., Noland, E., Hillier, C., Zisserman, A.: A short note on the Kinetics-700 human action dataset. arXiv preprint arXiv:1907.06987 (2019)

    2019 arXiv

  3. [11]

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

    Carreira, J., Zisserman, A.: Quo vadis, action recognition? a new model and the Kinetics dataset. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 6299–6308 (2017)

    2017

  4. [12]

    In: Proceedings of the 37th Interna- tional Conference on Machine Learning (ICML)

    Chen, T., Kornblith, S., Norouzi, M., Hinton, G.: A simple framework for con- trastive learning of visual representations. In: Proceedings of the 37th Interna- tional Conference on Machine Learning (ICML). Proceedings of Machine Learning Research, vol. 119, pp. 1597–1607. PMLR (2020)

    2020

  5. [13]

    In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Chen, X., He, K.: Exploring simple siamese representation learning. In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 15750–15758 (2021)

    2021

  6. [14]

    In: Proceedings of the IEEE/CVF International Conference on Com- puter Vision (ICCV)

    Chen, X., Xie, S., He, K.: An empirical study of training self-supervised vision transformers. In: Proceedings of the IEEE/CVF International Conference on Com- puter Vision (ICCV). pp. 9640–9649 (2021)

    2021

  7. [15]

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

    Danier, D., Aygün, M., Li, C., Bilen, H., Mac Aodha, O.: DepthCues: Evaluat- ing monocular depth perception in large vision models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 20049–20059 (2025)

    2025

  8. [16]

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

    Girdhar, R., El-Nouby, A., Singh, M., Alwala, K.V., Joulin, A., Misra, I.: Omni- MAE: Single model masked pretraining on images and videos. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 10406–10417 (2023)

    2023

  9. [17]

    Something Something

    Goyal, R., Kahou, S.E., Michalski, V., Materzynska, J., Westphal, S., Kim, H., Haenel, V., Fruend, I., Yianilos, P., Mueller-Freitag, M., Hoppe, F., Thurau, C., Bax, I., Memisevic, R.: The “Something Something” video database for learning and evaluating visual common sense. In...

    2017

  10. [18]

    In: Advances in Neural Information Processing Systems (NeurIPS)

    Grill, J.B., Strub, F., Altché, F., Tallec, C., Richemond, P., Buchatskaya, E., Do- ersch, C., Pires, B.A., Guo, Z., Azar, M.G., Piot, B., Kavukcuoglu, K., Munos, R., Valko, M.: Bootstrap your own latent: A new approach to self-supervised learning. In: Advances in Neural Infor...

    2020

  11. [19]

    He, K., Chen, X., Xie, S., Li, Y., Dollár, P., Girshick, R.: Masked autoencoders are scalablevisionlearners.In:ProceedingsoftheIEEE/CVFConferenceonComputer Vision and Pattern Recognition (CVPR). pp. 16000–16009 (2022)

    2022

  12. [20]

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

    He, K., Fan, H., Wu, Y., Xie, S., Girshick, R.: Momentum contrast for unsupervised visual representation learning. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 9729–9738 (2020)

    2020

  13. [21]

    In: Image and Video SSL Pretraining under Limited Resources 17 Proceedings of the European Conference on Computer Vision (ECCV)

    Hernandez, J., Villegas, R., Ordonez, V.: ViC-MAE: Self-supervised representa- tion learning from images and video with contrastive masked autoencoders. In: Image and Video SSL Pretraining under Limited Resources 17 Proceedings of the European Conference on Computer Vision (EC...

    2024 doi

  14. [22]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition (CVPR)

    Ke, B., Obukhov, A., Huang, S., Metzger, N., Daudt, R.C., Schindler, K.: Re- purposing diffusion-based image generators for monocular depth estimation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition (CVPR). pp. 9492–9502 (2024)

    2024

  15. [23]

    arXiv preprint arXiv:2511.13720 (2025)

    Li, T., He, K.: Back to basics: Let denoising generative models denoise. arXiv preprint arXiv:2511.13720 (2025)

    2025 arXiv

  16. [24]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2026)

    Norouzi, N., Zulfikar, I., Cavagnero, N., Kerssies, T., Leibe, B., Dubbelman, G., de Geus, D.: VidEoMT: Your ViT is Secretly Also a Video Segmentation Model. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2026)

    2026

  17. [25]

    Transactions on Ma- chine Learning Research (TMLR)2024(2024),https://openreview.net/forum? id=a68SUt6zFt

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H.V., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., Assran, M., Ballas, N., Galuba, W., Howes, R., Huang, P.Y., Li, S.W., Misra, I., Rabbat, M., Sharma, V., Syn- naeve, G., Xu, H., Jégou, H., Mairal...

    2024

  18. [26]

    Peebles,W.,Xie,S.:Scalablediffusionmodelswithtransformers.In:Proceedingsof the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 4195– 4205 (2023)

    2023

  19. [27]

    In: The Thirteenth International Conference on Learning Representations (2025),https://openreview.net/forum?id=Ha6RTeWMd0

    Ravi, N., Gabeur, V., Hu, Y.T., 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.Y., Girshick, R., Dollar, P., Feichtenhofer, C.: SAM 2: Segment anything in images and videos. In: The Thirteenth I...

    2025

  20. [28]

    In: International Conference on Learning Representations (2019),https: //openreview.net/forum?id=SkGuG2R5tm

    Sablayrolles, A., Douze, M., Schmid, C., Jégou, H.: Spreading vectors for similarity search. In: International Conference on Learning Representations (2019),https: //openreview.net/forum?id=SkGuG2R5tm

    2019

  21. [29]

    Neurocomputing568, 127063 (2024)

    Su, J., Ahmed, M., Lu, Y., Pan, S., Bo, W., Liu, Y.: RoFormer: Enhanced trans- former with Rotary Position Embedding. Neurocomputing568, 127063 (2024). https://doi.org/10.1016/j.neucom.2023.127063

    2024

  22. [30]

    In: Advances in Neu- ral Information Processing Systems (NeurIPS)

    Tong, Z., Song, Y., Wang, J., Wang, L.: VideoMAE: Masked autoencoders are data-efficient learners for self-supervised video pre-training. In: Advances in Neu- ral Information Processing Systems (NeurIPS). vol. 35, pp. 10078–10093 (2022). https://doi.org/10.52202/068431-0732

    2022 doi

  23. [31]

    https://www.tue.nl/en/research/research-groups/supercomputing-center

    TU/e Supercomputing Center: SPIKE High-Performance Computing Cluster. https://www.tue.nl/en/research/research-groups/supercomputing-center

  24. [32]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Vélez, P., Polanía, L.F., Yang, Y., Zhang, C., Kabra, R., Arnab, A., Sajjadi, M.S.M.: From image to video: An empirical study of diffusion representations. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 16948–16958 (2025)

    2025

  25. [33]

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

    Wang, R., Chen, D., Wu, Z., Chen, Y., Dai, X., Liu, M., Yuan, L., Jiang, Y.G.: Masked video distillation: Rethinking masked feature modeling for self-supervised video representation learning. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition...

    2023

  26. [34]

    Wang, Y., Li, K., Li, X., Yu, J., He, Y., Chen, G., Pei, B., Zheng, R., Wang, Z., Shi, Y., Jiang, T., Li, S., Xu, J., Zhang, H., Huang, Y., Qiao, Y., Wang, Y., Wang, L.: InternVideo2: Scaling foundation models for multimodal video understanding. 18 B.B. Englert and G. Dubbelma...

    2024 doi

  27. [35]

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

    Wei, C., Fan, H., Xie, S., Wu, C.Y., Yuille, A.L., Feichtenhofer, C.: Masked feature prediction for self-supervised visual pre-training. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 14668– 14678 (2022)

    2022

  28. [36]

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

    Xie, Z., Zhang, Z., Cao, Y., Lin, Y., Bao, J., Yao, Z., Dai, Q., Hu, H.: SimMIM: A simple framework for masked image modeling. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 9653–9663 (2022)

    2022

  29. [37]

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

    Zamir, A.R., Sax, A., Shen, W., Guibas, L.J., Malik, J., Savarese, S.: Taskonomy: Disentangling task transfer learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 3712–3722 (2018)

    2018

  30. [38]

    In: Proceedings of the 38th International Con- ference on Machine Learning (ICML)

    Zbontar, J., Jing, L., Misra, I., LeCun, Y., Deny, S.: Barlow twins: Self-supervised learning via redundancy reduction. In: Proceedings of the 38th International Con- ference on Machine Learning (ICML). Proceedings of Machine Learning Research, vol. 139, pp. 12310–12320. PMLR (2021)

    2021

  31. [39]

    Zhai, X., Puigcerver, J., Kolesnikov, A., Ruyssen, P., Riquelme, C., Lucic, M., Djo- longa, J., Susano Pinto, A., Neumann, M., Dosovitskiy, A., Beyer, L., Bachem, O., Tschannen, M., Michalski, M., Bousquet, O., Gelly, S., Houlsby, N.: The visual task adaptation benchmark (2020...

    2020

  32. [40]

    In: International Conference on Learning Representations (ICLR) (2022),https://openreview.net/forum?id=ydopy-e6Dg

    Zhou, J., Wei, C., Wang, H., Shen, W., Xie, C., Yuille, A.L., Kong, T.: iBOT: Image BERT pre-training with online tokenizer. In: International Conference on Learning Representations (ICLR) (2022),https://openreview.net/forum?id=ydopy-e6Dg

    2022

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.