Pith. sign in

REVIEW 3 major objections 6 minor 80 references

Pre-training for Action Recognition with Automatically Generated Fractal Datasets

T0 review · 3 major / 6 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 Pre-training action recognition models on automatically generated fractal videos reaches accuracy close to Kinetics pre-training and surpasses it on two of six benchmarks, with no real video data required.

desk verdict A genuinely new extension of fractal-image pre-training to video, with a broad and honest empirical study, but the 'superior to Kinetics' claim rests on an uncontrolled baseline and no seed variance. read the letter →

arxiv 2411.17584 v1 pith:7TEDB6QS submitted 2024-11-26 cs.CV

Davyd Svyezhentsev , George Retsinas , Petros Maragos This is my paper · ORCID
classification cs.CV
keywords fractalgeometrysyntheticdataactionrecognitionvideopre-trainingdomainadaptationiteratedfunctionsystemsself-supervisedlearningtemporalshiftmodule
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 tries to establish that action recognition models can be pre-trained on synthetic videos generated automatically from fractal geometry, replacing large real datasets such as Kinetics with all their collection, annotation, copyright, and privacy costs. It builds a pipeline that renders large corpora of unique fractal clips from iterated function systems, animates them by interpolating decomposed transformation parameters, and narrows the domain gap to real video with augmentations that mimic human motion, backgrounds, and camera movement. After fine-tuning a TSM/ResNet-50 on six action recognition benchmarks, the fractal-pretrained model reaches accuracy close to Kinetics pre-training and actually surpasses it on DIVING48 and VOLLEYBALL, the two datasets with the smallest background variance. If the claims hold, video pre-training no longer requires collecting or labeling real footage; the main levers are the diversity, size, and resolution of the synthetic corpus and how well it mirrors the target domain.

What carries the argument

The central object is the iterated function system (IFS) fractal: a set of contractive affine maps $F_i(x) = A_i x + b_i$ whose attractor is rendered as a grayscale image by the chaos game. To make videos, each $A_i$ is decomposed as $A = R_\theta \Sigma R_\phi D$ (rotation by $\theta$, scaling by singular values, rotation by $\phi$, and a reflection $D$ with $\pm 1$ entries), the pieces are linearly interpolated between two keyframes, and recomposed frame by frame, a scheme that avoids the sparseness that naive interpolation of the raw parameters produces. Around this core sit three supporting mechanisms: nonlinear variations from the fractal-flame algorithm that multiply the diversity of rendered shapes; a mutation rule $\tilde{W}_c = m_a \odot W_c + m_b$ that turns one sampled fractal into an entire category of related clips for supervised pre-training; and a battery of online augmentations, including nonlinear motion curves, static or dynamic random backgrounds, foreground scaling and placement, group clones, perspective, foreground/background/camera displacement, zoom, and shake, that emulate the structural and motion properties of real videos.

What would settle it

Fine-tune the paper's selected configuration on a genuinely held-out test split of HMDB51 and UCF101, or on fresh random splits repeated over several seeds, always against the same Kinetics checkpoint. If the fractal-pretrained model's accuracy drops materially relative to Kinetics, for instance an HMDB51 gap well beyond the reported 3.6 points or a reversal of the DIVING48 and VOLLEYBALL advantages, then the central claim fails.

Watch

Extended reading notes

Core claim

Pre-training a TSM/ResNet-50 on 400K automatically generated fractal video clips, with the proposed domain adaptations, then fine-tuning on real benchmarks, reaches top-1 accuracy of 66.5/90.8/41.2/59.9/87.6/78.2 on HMDB51, UCF101, DIVING48, EGTEA, VOLLEYBALL, and YUP++; Kinetics pre-training at the same 224-pixel resolution reaches 70.1/95.3/40.9/64.4/84.8/86.9. Synthetic pre-training therefore beats Kinetics on DIVING48 and VOLLEYBALL, the datasets with the least background variance, and trails it on the other four. The paper further asserts that the diversity of synthetic videos is the key factor for strong representations, boosting results on every benchmark regardless of the downstream domain, and that supervised pre-training on arbitrary, non-interpretable categories transfers better than the self-supervised objectives MoCoV2, SimCLR, and BYOL under matched resources. It also reports that fractal pre-training beats ImageNet initialization on Something-Something V2 for both TSM and I3D, evidence that the synthetic clips carry usable motion information that static-image pre-training lacks.

Load-bearing premise

The load-bearing premise is that validation accuracy on HMDB51 and UCF101, the same sets used to decide which domain adaptations to keep, is an unbiased estimate of generalization on those datasets. If that selection overfits the benchmarks, the headline match with Kinetics may not hold on unseen test data.

Editorial extensions

If this is right

  • Video pre-training no longer requires collecting or labeling real footage: fractal clips are generated automatically, sidestepping collection effort, annotation cost, copyright, bias, and privacy concerns.
  • On the two benchmarks with the least background variation, DIVING48 and VOLLEYBALL, fractal pre-training beats Kinetics (41.2 vs 40.9 and 87.6 vs 84.8) using 400K synthetic clips and 25 training epochs rather than 250K real clips and 100 epochs.
  • Supervised pre-training on arbitrarily generated, semantically meaningless categories transfers better than the self-supervised objectives MoCoV2, SimCLR, and BYOL under the same budget, making the cheap supervised setting the practical choice.
  • Diversity of the synthetic corpus is the one factor that improves every downstream benchmark, while matching the target domain's background, motion, and camera properties decides where transfer succeeds and where it does not.
  • Fractal pre-training also beats ImageNet initialization on Something-Something V2 for both TSM (59.7 vs 58.8) and I3D (52.6 vs 51.2), showing the synthetic clips carry motion information that static-image pre-training cannot provide.

Reading between the lines

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

  • The paper's explanation for beating Kinetics, that DIVING48 and VOLLEYBALL have unusually low background variance, is testable: generate synthetic datasets with deliberately varied background diversity and check whether the advantage over Kinetics tracks background variance causally.
  • The success of supervised pre-training on meaningless categories suggests that the category structure itself, not its semantic content, is what transfers; an ablation that holds dataset size fixed while varying the number of categories could separate the contribution of label structure from raw data diversity.
  • The paper's own error analysis points to a concrete next step it leaves untried: construct synthetic categories whose labels depend on a small region of pixels, which should directly target EGTEA and the fine-detail classes of HMDB51 and UCF101 where fractal pre-training currently loses ground.
  • Because accuracy rose monotonically with both classes and instances per class, scaling the synthetic corpus well beyond 400K clips at 224 resolution is the natural test of whether the remaining gap to Kinetics on HMDB51, UCF101, EGTEA, and YUP++ closes or plateaus, and the paper itself expects it to shrink but not vanish.
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 / 6 minor

Summary. This paper proposes to pre-train action-recognition models on automatically generated synthetic videos based on fractal geometry. The authors describe a pipeline that generates short video clips by interpolating parameters of iterated function systems, construct pseudo-classes via parameter mutation, and introduce a set of hand-designed domain adaptations (nonlinear motion, background mixing, scaling, displacement, zoom, shake, perspective, group activity) intended to emulate properties of real videos. They compare supervised and self-supervised pre-training objectives, ablate dataset size and resolution, and evaluate by fine-tuning TSM/ResNet-50 on HMDB51, UCF101, DIVING48, EGTEA GAZE+, VOLLEYBALL, YUP++, and Something-Something V2. The central empirical claim is that fractal pre-training comes close to Kinetics pre-training and is even superior on DIVING48 and VOLLEYBALL, while using far less computation and no real data.

Significance. If the central claim is substantiated, the paper makes a useful contribution: a fully automatic, privacy- and copyright-free source of pre-training videos that approaches the performance of large real-video pre-training on several benchmarks. The paper is strong in breadth of ablations (domain adaptations, alternative synthetic generators, training objectives, dataset size, resolution) and in honest reporting of limitations, and the authors release code and samples. The main scientific value lies in the guidelines for synthetic video pre-training, especially the evidence that synthetic diversity and rough emulation of real-video motion properties help transfer. However, the headline comparison to Kinetics is currently weakened by evaluation-protocol confounds that need to be addressed before the claimed superiority on DIVING48 and VOLLEYBALL can be taken at face value.

major comments (3)
  1. [§3.2 and Table 6] The central comparison against Kinetics is based on a configuration that was selected on the same validation sets used for reporting. Section 3.2 states that a domain-adaptation technique is discarded if improvement is not achieved on HMDB51 and UCF101, and that the configuration with highest accuracy on those two sets is retained. The final numbers in Table 6 for HMDB51 and UCF101 are therefore validation-set-tuned, and the same shared configuration is used for the other four datasets. Since there is no held-out test split and no repeated-seed variance, the reported 0.3-point margin over Kinetics on DIVING48 (41.2 vs 40.9) and the 2.8-point margin on VOLLEYBALL could be inflated by selection noise. Please report results with at least three random seeds, use a separate selection split (or nested validation), or otherwise quantify how much of the reported ordering survives configuration selection.
  2. [§3.1, Appendix A, Table 6] The Kinetics baseline is not an apples-to-apples comparison. The Kinetics checkpoint from [50] was pre-trained for 100 epochs on 250K real videos, whereas the fractal model is pre-trained for 25 epochs on 400K synthetic videos, and fine-tuning after Kinetics uses different hyperparameters (Table 8) than the protocol in Section 3.1. Thus Table 6 conflates the source-domain difference with differences in pre-training data size, training schedule, and fine-tuning procedure. Please fine-tune the same Kinetics checkpoint using the exact Section 3.1 fine-tuning protocol, and ideally also compare under matched pre-training data size and schedule, so that the contrast isolates real versus synthetic pre-training.
  3. [Tables 1–7 (all experiments)] Every reported accuracy in the paper appears to come from a single run, with no error bars, confidence intervals, or seed variance. For the headline claims this is not a cosmetic issue: on a dataset of DIVING48's size, a 0.3-point difference is reasonably within run-to-run variation, and even the 2.8-point VOLLEYBALL gap needs a variance estimate to be credible. Please add repeated-seed statistics at least for Table 6 and, if feasible, for the main ablation tables, so readers can distinguish systematic effects from optimization noise.
minor comments (6)
  1. [§2.3, camera shake equation] The subscript is missing in the displacement formula: the equation should read d_t = sum_i (1/i) sin(2π f_i t + φ) + η_t, not "dt =".
  2. [§3.1] The learning-rate scaling formula appears as "ltrue = bs bsbase lr"; it should be written as lr_true = (bs / bs_base) · lr for readability.
  3. [Figure 2 caption] The caption labels "Satisfactory Nonlinear Sparse Dense" are ambiguous; please clarify which images correspond to which qualitative category.
  4. [Table 1 caption] The caption refers to green, red, orange, blue, and yellow colors, but tables in this version appear to be grayscale; please add textual markers or symbols so the distinctions survive printing.
  5. [§3.2, Experiment 1] There is a typo: "One the contrary" should be "On the contrary."
  6. [§5] The paper's first limitation bullet already concedes that synthetic pre-training lags behind Kinetics on the majority of benchmarks; the abstract and Section 5 claim of being "even superior on a portion" should be scoped explicitly to the two low-background-variance datasets so that the conclusion matches the evidence.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the paper's claims are empirical benchmark comparisons, not derivations, and no load-bearing step reduces to its own inputs by construction.

full rationale

This paper does not contain a derivation chain of the kind the circularity taxonomy targets. The synthetic video generation pipeline (Algorithm 1), the domain-adaptation augmentations (Sec. 2.3), and the training objectives (Sec. 2.5) are constructive procedures, and downstream results are obtained by fine-tuning on real, externally labeled benchmarks (HMDB51, UCF101, DIVING48, EGTEA GAZE+, VOLLEYBALL, YUP++, SSv2). No fitted parameter is renamed as a prediction: the reported accuracies are measured validation-set accuracies after fine-tuning, and the Kinetics comparison uses an independent off-the-shelf checkpoint from prior work (Sec. 3.2, Experiment 5). The configuration-selection rule in Sec. 3.2 ('After each experiment, we retain the configuration with highest accuracy on HMDB51 and UCF101') is a benchmark-selection concern, not an equation-level circularity; it does not make any reported number equal to an input by construction, and the claimed superiority on DIVING48 and VOLLEYBALL rests on datasets that were not used for that selection. Citations to prior work (e.g., fractal pre-training [42], improved fractal sampling [2], fractal flames [20]) are external, and the few self-citations by co-author Maragos concern fractal signal analysis background rather than the paper's load-bearing claims. The absence of repeated-seed variance and the differing Kinetics fine-tuning protocol are robustness limitations, but they do not constitute circularity under the required standard.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The method's success rests on many hand-picked generation parameters and augmentations, and the most influential choices (which domain adaptations to keep) are tuned on the evaluation datasets. No formal derivation selects these values.

free parameters (6)
  • Number of synthetic pre-training videos = 100K to 400K
    Dataset size swept in Experiment 4; final best model uses 400K videos (Table 5).
  • Number of classes and instances per class = 500 classes x 200 instances for 100K; final best uses 1000 classes x 400 instances
    Chosen by hand and swept in Experiments 4A/4B; downstream accuracy increases with both.
  • Domain augmentation probabilities = p=0.3 for most, 1.0 for Background and Scale, 0.8 for Perspective, 0.15 for Group
    Set heuristically in Sec. 3.1; the keep/discard decisions in Table 1 are the main tuning lever.
  • Mutation noise bounds = ma in [-0.35,0.35], mb in U(-0.2,0.2)
    Define the intra-class variation for supervised category construction (Sec. 2.3.1) and are not justified beyond visual inspection.
  • Foreground scale ranges = smin=0.3, smax=1.0 (single); smin=0.2, smax=0.7 (group)
    Chosen by hand in Sec. 2.3 to mimic real-world object sizes; not optimized with validation.
  • Synthetic video resolution and clip length = 256x256 render, 112 or 224 input, 18-20 frames
    Standard choices; resolution sweep in Experiment 5 shows monotonic improvement.
assumptions (5)
  • standard math IFS attractor existence and chaos game convergence for contractive affine maps (Barnsley 1993)
    Relied upon in Sec. 2.1.1 to generate fractal images.
  • domain assumption The empirical bound a = sum(sigma_i1 + 2sigma_i2) in [0.5(5+N), 0.5(6+N)] from Anderson and Farrell (2022) yields non-degenerate fractals
    Adopted in Sec. 2.1.1 to sample 'well-behaved' attractors; the bound is fitted to visual quality in the cited paper.
  • ad hoc to paper The hand-designed augmentations (background mixing, camera displacement, zoom, shake, etc.) emulate the key properties of real videos enough to narrow the domain gap
    The central premise of Sec. 2.3; the specific set and parameter ranges are justified only by downstream validation results.
  • domain assumption Supervised pre-training on randomly generated, semantically meaningless fractal categories transfers to real action recognition
    The paper shows this empirically (Table 3) but does not explain why learning arbitrary shape categories should produce action-relevant features.
  • domain assumption TSM/ResNet-50 is a representative architecture for video pre-training
    All main experiments use this single architecture; only SSv2 experiments add I3D (Table 7).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Pre-training for Action Recognition with Automatically Generated Fractal Datasets." pith.science (2026). https://pith.science/paper/7TEDB6QS

@misc{pith2026241117584,
  author       = {Pith},
  title        = {Pith review of: Pre-training for Action Recognition with Automatically Generated Fractal Datasets},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7TEDB6QS}},
  note         = {Machine review of arXiv:2411.17584}
}
read the original abstract

In recent years, interest in synthetic data has grown, particularly in the context of pre-training the image modality to support a range of computer vision tasks, including object classification, medical imaging etc. Previous work has demonstrated that synthetic samples, automatically produced by various generative processes, can replace real counterparts and yield strong visual representations. This approach resolves issues associated with real data such as collection and labeling costs, copyright and privacy. We extend this trend to the video domain applying it to the task of action recognition. Employing fractal geometry, we present methods to automatically produce large-scale datasets of short synthetic video clips, which can be utilized for pre-training neural models. The generated video clips are characterized by notable variety, stemmed by the innate ability of fractals to generate complex multi-scale structures. To narrow the domain gap, we further identify key properties of real videos and carefully emulate them during pre-training. Through thorough ablations, we determine the attributes that strengthen downstream results and offer general guidelines for pre-training with synthetic videos. The proposed approach is evaluated by fine-tuning pre-trained models on established action recognition datasets HMDB51 and UCF101 as well as four other video benchmarks related to group action recognition, fine-grained action recognition and dynamic scenes. Compared to standard Kinetics pre-training, our reported results come close and are even superior on a portion of downstream datasets. Code and samples of synthetic videos are available at https://github.com/davidsvy/fractal_video .

Figures

Figures reproduced from arXiv: 2411.17584 by the authors.

Figure 1
Figure 1. Overview of the proposed approach. Aiming to pre-train neural models, we utilize fractal geometry and [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Examples of rendered 2D IFS attractors. A subset of linear samples exhibits unsatisfactory geometry, being [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. An example of the the proposed animation method compared to naive interpolation (Sec. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Examples of the proposed nonlinear interpola [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Examples of the proposed domain adaptation [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Examples of the proposed mutation scheme. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Examples of alternative synthetic videos. Com [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Frames taken from the evaluated downstream datasets. Note the variance of the displayed background in [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Frames from misclassified videos. Blue color indicates ground truth, whereas red color indicates the model’s incorrect prediction. In such videos the label is often determined by subtle details that cover a small percentage of the overall pixels. As such, the model fai…
Figure 10
Figure 10. Figure 10: Examples of videos produced with Algorithm [PITH_FULL_IMAGE:figures/full_fig_p021_10.png]
Figure 11
Figure 11. Figure 11: Examples of videos produced with Algorithm [PITH_FULL_IMAGE:figures/full_fig_p022_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

80 extracted references · 71 canonical work pages

  1. [50]

    In: Proceedings of the IEEE International Conference on Computer Vision (ICCV) (2019) 9, 13, 20

    Lin, J., Gan, C., Han, S.: Tsm: Temporal shift mod- ule for efficient video understanding. In: Proceedings of the IEEE International Conference on Computer Vision (ICCV) (2019) 9, 13, 20

    work page 2019

  2. [1]

    In: arXiv preprint arXiv:1609.08675 (2016) 16

    Abu-El-Haija, S., Kothari, N., Lee, J., Natsev, P., Toderici, G., Varadarajan, B., Vijayanarasimhan, S.: Youtube-8m: A large-scale video classification bench- mark. In: arXiv preprint arXiv:1609.08675 (2016) 16

    arXiv 2016

  3. [2]

    In: Proceedings of the IEEE Winter Conference on Ap- plications of Computer Vision (W ACV) (2022) 1, 3, 4, 5, 14

    Anderson, C., Farrell, R.: Improving fractal pre-training. In: Proceedings of the IEEE Winter Conference on Ap- plications of Computer Vision (W ACV) (2022) 1, 3, 4, 5, 14

    work page 2022

  4. [3]

    In: Fractal and Fractional, vol

    Anguera, J., And´ ujar, A., Jayasinghe, J., Chakravarthy, V.V.S.S.S., Chowdary, P.S.R., Pijoan, J.L., Ali, T., Cat- Pre-training for Action Recognition with Automatically Generated Fractal Datasets 17 tani, C.: Fractal antennas: An historical perspective. In: Fractal and Fractional, vol. 4, no. 1 (2020) 16

    work page 2020

  5. [4]

    In: Proceedings of the Interna- tional Conference on Neural Information Processing Sys- tems (NeurIPS) Datasets and Benchmarks Track (2021) 1

    Asano, Y., Rupprecht, C., Zisserman, A., Vedaldi, A.: Pass: An imagenet replacement for self-supervised pre- training without humans. In: Proceedings of the Interna- tional Conference on Neural Information Processing Sys- tems (NeurIPS) Datasets and Benchmarks Track (2021) 1

    work page 2021

  6. [5]

    In: Pro- ceedings of the International Conference on Neural Infor- mation Processing Systems (NeurIPS) (2021) 1, 4, 5, 8, 11, 15, 16

    Baradad Jurjo, M., Wulff, J., Wang, T., Isola, P., Tor- ralba, A.: Learning to see by looking at noise. In: Pro- ceedings of the International Conference on Neural Infor- mation Processing Systems (NeurIPS) (2021) 1, 4, 5, 8, 11, 15, 16

    work page 2021

  7. [6]

    Morgan Kaufmann (1993) 1, 2, 3

    Barnsley, M.F.: Fractals Everywhere. Morgan Kaufmann (1993) 1, 2, 3

    work page 1993

  8. [7]

    Birhane, A., Prabhu, V.U.: Large image datasets: A pyrrhic win for computer vision? In: Proceedings of the IEEE Winter Conference on Applications of Computer Vision (W ACV) (2021) 1

    work page 2021

Show all 80 references
  1. [8]

    In: Proceedings of the Conference on Fairness, Account- ability and Transparency (2018) 1

    Buolamwini, J., Gebru, T.: Gender shades: Intersectional accuracy disparities in commercial gender classification. In: Proceedings of the Conference on Fairness, Account- ability and Transparency (2018) 1

    2018

  2. [9]

    In: Proceedings of the Graphics Interface Confer- ence (1997) 4

    Burch, B., Hart, J.C.: Linear fractal shape interpola- tion. In: Proceedings of the Graphics Interface Confer- ence (1997) 4

    1997

  3. [10]

    In: arXiv preprint arXiv:1907.06987 (2019) 2, 8, 13, 16

    Carreira, J., Noland, E., Hillier, C., Zisserman, A.: A short note on the kinetics-700 human action dataset. In: arXiv preprint arXiv:1907.06987 (2019) 2, 8, 13, 16

    2019 arXiv

  4. [11]

    In: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017) 13, 16

    Carreira, J., Zisserman, A.: Quo vadis, action recogni- tion? a new model and the kinetics dataset. In: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017) 13, 16

    2017

  5. [12]

    In: Proceedings of the International Conference on Machine Learning (ICML) (2020) 8, 12

    Chen, T., Kornblith, S., Norouzi, M., Hinton, G.: A sim- ple framework for contrastive learning of visual represen- tations. In: Proceedings of the International Conference on Machine Learning (ICML) (2020) 8, 12

    2020

  6. [13]

    In: arXiv preprint arXiv:2003.04297 (2020) 8, 11, 12, 13

    Chen, X., Fan, H., Girshick, R., He, K.: Improved base- lines with momentum contrastive learning. In: arXiv preprint arXiv:2003.04297 (2020) 8, 11, 12, 13

    2020 arXiv

  7. [14]

    Cole, E., Yang, X., Wilber, K., Mac Aodha, O., Belongie, S.: When does contrastive visual representation learning work? In: Proceedings of the IEEE Conference on Com- puter Vision and Pattern Recognition (CVPR) (2022) 4, 11, 12

    2022

  8. [15]

    In: Journal of Biomechan- ics, vol

    Costabal, F., Hurtado, D., Kuhl, E.: Generating purkinje networks in the human heart. In: Journal of Biomechan- ics, vol. 49, pp. 2455–2465 (2015) 16

    2015

  9. [16]

    In: Proceedings of the International Conference on Neural Information Processing Systems (NeurIPS) (2020) 10

    Cubuk, E.D., Zoph, B., Shlens, J., Le, Q.: Randaug- ment: Practical automated data augmentation with a re- duced search space. In: Proceedings of the International Conference on Neural Information Processing Systems (NeurIPS) (2020) 10

    2020

  10. [17]

    In: IEEE Transactions on Signal Process- ing, vol

    Dimakis, A., Maragos, P.: Phase-modulated resonances modeled as self-similar processes with application to tur- bulent sounds. In: IEEE Transactions on Signal Process- ing, vol. 53, no. 11, pp. 4261–4272 (2005) 16

    2005

  11. [18]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2022) 6

    Ding, S., Li, M., Yang, T., Qian, R., Xu, H., Chen, Q., Wang, J., Xiong, H.: Motion-aware contrastive video rep- resentation learning via foreground-background merging. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2022) 6

    2022

  12. [19]

    In: Proceedings of the European conference on Applications of Evolutionary Computing (2005) 16

    Draves, S.: The electric sheep screen-saver: A case study in aesthetic evolution. In: Proceedings of the European conference on Applications of Evolutionary Computing (2005) 16

    2005

  13. [20]

    (2008) (2008) 4, 6, 16

    Draves, S., Reckase, E.: The fractal flame algorithm. (2008) (2008) 4, 6, 16

    2008

  14. [21]

    Fan, L., Chen, K., Krishnan, D., Katabi, D., Isola, P., Tian, Y.: Scaling laws of synthetic images for model train- ing ... for now. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2024) 15

    2024

  15. [22]

    Academic Press Pro- fessional, Inc

    Farin, G.: Curves and Surfaces for Computer Aided Ge- ometric Design: A Practical Guide. Academic Press Pro- fessional, Inc. (1988) 15

    1988

  16. [23]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2021) 16

    Feichtenhofer, C., Fan, H., Xiong, B., Girshick, R., He, K.: A large-scale study on unsupervised spatiotemporal representation learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2021) 16

    2021

  17. [24]

    In: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017) 9

    Feichtenhofer, C., Pinz, A., Wildes, R.P.: Temporal resid- ual networks for dynamic scene recognition. In: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017) 9

    2017

  18. [25]

    In: Proceedings of the IEEE Conference on Com- puter Vision and Pattern Recognition (CVPR) (2019) 1

    Ghadiyaram, D., Tran, D., Mahajan, D.: Large-scale weakly-supervised pre-training for video action recogni- tion. In: Proceedings of the IEEE Conference on Com- puter Vision and Pattern Recognition (CVPR) (2019) 1

    2019

  19. [26]

    In: Proceedings of the In- ternational Conference on Neural Information Processing Systems (NeurIPS) (2014) 15

    Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., Bengio, Y.: Generative adversarial nets. In: Proceedings of the In- ternational Conference on Neural Information Processing Systems (NeurIPS) (2014) 15

    2014

  20. [27]

    In: arXiv preprint arXiv:2103.01988 (2021) 1

    Goyal, P., Caron, M., Lefaudeux, B., Xu, M., Wang, P., Pai, V., Singh, M., Liptchinsky, V., Misra, I., Joulin, A., Bojanowski, P.: Self-supervised pretraining of visual fea- tures in the wild. In: arXiv preprint arXiv:2103.01988 (2021) 1

    2021 arXiv

  21. [28]

    In: Proceedings of the IEEE International Conference on Computer Vision (ICCV) (2017) 13

    Goyal, R., Ebrahimi Kahou, S., Michalski, V., Materzyn- ska, 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 com- mo...

    2017

  22. [29]

    In: Proceedings of the International Conference on Neural Information Process- ing Systems (NeurIPS) (2020) 8, 12

    Grill, J.B., Strub, F., Altch´ e, F., Tallec, C., Richemond, P., Buchatskaya, E., Doersch, C., Avila Pires, B., Guo, Z., Gheshlaghi Azar, M., Piot, B., kavukcuoglu, k., Munos, R., Valko, M.: Bootstrap your own latent - a new ap- proach to self-supervised learning. In: Proceedi...

    2020

  23. [30]

    Hara, K., Kataoka, H., Satoh, Y.: Can spatiotemporal 3d cnns retrace the history of 2d cnns and imagenet? In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2018) 16

    2018

  24. [31]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2016) 9

    He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learn- ing for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2016) 9

    2016

  25. [32]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2015) 16

    Heilbron, F.C., Escorcia, V., Ghanem, B., Niebles, J.C.: Activitynet: A large-scale video benchmark for human activity understanding. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2015) 16

    2015

  26. [33]

    In: Proceedings of the International Conference on Neural Information Processing Systems (NeurIPS) (2020) 15

    Ho, J., Jain, A., Abbeel, P.: Denoising diffusion prob- abilistic models. In: Proceedings of the International Conference on Neural Information Processing Systems (NeurIPS) (2020) 15

    2020

  27. [34]

    In: Proceedings 18 Davyd Svyezhentsev et al

    Ho, J., Salimans, T., Gritsenko, A., Chan, W., Norouzi, M., Fleet, D.J.: Video diffusion models. In: Proceedings 18 Davyd Svyezhentsev et al. of the International Conference on Neural Information Processing Systems (NeurIPS) (2022) 15

    2022

  28. [35]

    In: Indiana University Mathematics Journal, vol

    HUTCHINSON, J.E.: Fractals and self similarity. In: Indiana University Mathematics Journal, vol. 30, no. 5, pp. 713–747 (1981) 3

    1981

  29. [36]

    In: Proceedings of the IEEE Con- ference on Computer Vision and Pattern Recognition (CVPR) (2016) 9

    Ibrahim, M.S., Muralidharan, S., Deng, Z., Vahdat, A., Mori, G.: A hierarchical deep temporal model for group activity recognition. In: Proceedings of the IEEE Con- ference on Computer Vision and Pattern Recognition (CVPR) (2016) 9

    2016

  30. [37]

    In: IF AC Proceedings Volumes, vol

    Ionescu, C., Oustaloup, A., Levron, F., Melchior, P., Sabatier, J., De Keyser, R.: A model of the lungs based on fractal geometrical and structural properties. In: IF AC Proceedings Volumes, vol. 42, no. 10, pp. 994–999 (2009) 16

    2009

  31. [38]

    In: IEEE transactions on image processing, vol

    Jacquin, A.E.: Image coding based on a fractal theory of iterated contractive image transformations. In: IEEE transactions on image processing, vol. 1 1, pp. 18–30 (1992) 16

    1992

  32. [39]

    In: Proceedings of the IEEE Winter Conference on Applications of Computer Vision (W ACV) (2022) 8, 15

    Kataoka, H., Hara, K., Hayashi, R., Yamagata, E., In- oue, N.: Spatiotemporal initialization for 3d cnns with generated motion patterns. In: Proceedings of the IEEE Winter Conference on Applications of Computer Vision (W ACV) (2022) 8, 15

    2022

  33. [40]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2022) 14

    Kataoka, H., Hayamizu, R., Yamada, R., Nakashima, K., Takashima, S., Zhang, X., Martinez-Noriega, E.J., Inoue, N., Yokota, R.: Replacing labeled real-image datasets with auto-generated contours. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition ...

    2022

  34. [41]

    In: Proceedings of the IEEE International Conference on Computer Vision (ICCV) Workshops (2021) 1

    Kataoka, H., Matsumoto, A., Yamada, R., Satoh, Y., Ya- magata, E., Inoue, N.: Formula-driven supervised learn- ing with recursive tiling patterns. In: Proceedings of the IEEE International Conference on Computer Vision (ICCV) Workshops (2021) 1

    2021

  35. [42]

    In: Proceedings of the Asian Conference on Computer Vision (ACCV) (2020) 1, 2, 3, 7, 14

    Kataoka, H., Okayasu, K., Matsumoto, A., Yamagata, E., Yamada, R., Inoue, N., Nakamura, A., Satoh, Y.: Pre- training without natural images. In: Proceedings of the Asian Conference on Computer Vision (ACCV) (2020) 1, 2, 3, 7, 14

    2020

  36. [43]

    In: arXiv preprint arXiv:1705.06950 (2017) 2, 8, 13, 16

    Kay, W., Carreira, J., Simonyan, K., Zhang, B., Hillier, C., Vijayanarasimhan, S., Viola, F., Green, T., Back, T., Natsev, P., Suleyman, M., Zisserman, A.: The ki- netics human action video dataset. In: arXiv preprint arXiv:1705.06950 (2017) 2, 8, 13, 16

    2017 arXiv

  37. [44]

    In: IEEE Transactions on Speech and Audio Processing, vol

    Kokkinos, I., Maragos, P.: Nonlinear speech analysis us- ing models for chaotic systems. In: IEEE Transactions on Speech and Audio Processing, vol. 13, no. 6, pp. 1098– 1109 (2005) 16

    2005

  38. [45]

    In: Proceedings of the IEEE Interna- tional Conference on Computer Vision (ICCV) (2021) 4, 11, 12

    Kotar, K., Ilharco, G., Schmidt, L., Ehsani, K., Mottaghi, R.: Contrasting contrastive self-supervised representation learning pipelines. In: Proceedings of the IEEE Interna- tional Conference on Computer Vision (ICCV) (2021) 4, 11, 12

    2021

  39. [46]

    In: Proceedings of the IEEE International Conference on Computer Vision (ICCV) (2011) 5, 9, 16

    Kuehne, H., Jhuang, H., Garrote, E., Poggio, T., Serre, T.: Hmdb: A large video database for human motion recognition. In: Proceedings of the IEEE International Conference on Computer Vision (ICCV) (2011) 5, 9, 16

    2011

  40. [47]

    In: International Journal of Computer Vision (IJCV), vol

    Lee, A.B., Mumford, D., Huang, J.: Occlusion models for natural images: A statistical study of a scale-invariant dead leaves model. In: International Journal of Computer Vision (IJCV), vol. 41, pp. 35–59 (2004) 8

    2004

  41. [48]

    In: Proceedings of the European Conference on Computer Vision (ECCV) (2018) 9

    Li, Y., Li, Y., Vasconcelos, N.: Resound: Towards action recognition without representation bias. In: Proceedings of the European Conference on Computer Vision (ECCV) (2018) 9

    2018

  42. [49]

    In: Proceedings of the European Conference on Computer Vision (ECCV) (2018) 9

    Li, Y., Liu, M., Rehg, J.M.: In the eye of beholder: Joint learning of gaze and actions in first person video. In: Proceedings of the European Conference on Computer Vision (ECCV) (2018) 9

    2018

  43. [51]

    In: Proceedings of the Interna- tional Conference on Learning Representations (ICLR) (2017) 10

    Loshchilov, I., Hutter, F.: SGDR: Stochastic gradient de- scent with warm restarts. In: Proceedings of the Interna- tional Conference on Learning Representations (ICLR) (2017) 10

    2017

  44. [52]

    In: Proceedings of the International Confer- ence on Learning Representations (ICLR) (2019) 10

    Loshchilov, I., Hutter, F.: Decoupled weight decay regu- larization. In: Proceedings of the International Confer- ence on Learning Representations (ICLR) (2019) 10

    2019

  45. [53]

    In: Advances in Electronics and Electron Physics, vol

    Maragos, P.: Fractal signal analysis using mathematical morphology. In: Advances in Electronics and Electron Physics, vol. 88, pp. 199–246 (1994) 16

    1994

  46. [54]

    In: The Journal of the Acous- tical Society of America, vol

    Maragos, P., Potamianos, A.: Fractal dimensions of speech sounds: Computation and application to auto- matic speech recognition. In: The Journal of the Acous- tical Society of America, vol. 105, pp. 1925–32 (1999) 16

    1999

  47. [55]

    In: arXiv preprint arXiv:2405.03150 (2024) 15

    Melnik, A., Ljubljanac, M., Lu, C., Yan, Q., Ren, W., Ritter, H.: Video diffusion models: A survey. In: arXiv preprint arXiv:2405.03150 (2024) 15

    2024 arXiv

  48. [56]

    In: IEEE Transactions on Pattern Analysis and Machine Intelli- gence, vol

    Monfort, M., Andonian, A., Zhou, B., Ramakrishnan, K., Bargal, S.A., Yan, T., Brown, L., Fan, Q., Gutfreund, D., Vondrick, C., Oliva, A.: Moments in time dataset: One million videos for event understanding. In: IEEE Transactions on Pattern Analysis and Machine Intelli- gence, ...

    2020

  49. [57]

    Nakashima, K., Kataoka, H., Matsumoto, A., Iwata, K., Inoue, N., Satoh, Y.: Can vision transformers learn with- out natural images? In: Proceedings of the AAAI Con- ference on Artificial Intelligence (2022) 14

    2022

  50. [58]

    In: Proceedings of the Annual Conference on Computer Graphics and Interac- tive Techniques (SIGGRAPH) (1985) 8, 15

    Perlin, K.: An image synthesizer. In: Proceedings of the Annual Conference on Computer Graphics and Interac- tive Techniques (SIGGRAPH) (1985) 8, 15

    1985

  51. [59]

    In: Proceedings of the An- nual Conference on Computer Graphics and Interactive Techniques (SIGGRAPH) (2002) 8, 15

    Perlin, K.: Improving noise. In: Proceedings of the An- nual Conference on Computer Graphics and Interactive Techniques (SIGGRAPH) (2002) 8, 15

    2002

  52. [60]

    In: Speech Communication, vol

    Pitsikalis, V., Maragos, P.: Analysis and classification of speech signals by generalized fractal dimension features. In: Speech Communication, vol. 51, no. 12, pp. 1206–1223 (2009) 16

    2009

  53. [61]

    In: Proceed- ings of the Visual Communications and Image Processing (VCIP) (2013) 7

    Qu, H., Song, L., Xue, G.: Shaking video synthesis for video stabilization performance assessment. In: Proceed- ings of the Visual Communications and Image Processing (VCIP) (2013) 7

    2013

  54. [62]

    In: Vision Research, vol

    Ruderman, D.L.: Origins of scaling in natural images. In: Vision Research, vol. 37, no. 23, pp. 3385–3398 (1997) 8

    1997

  55. [63]

    In: International Journal of Computer Vision (IJCV), vol

    Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bern- stein, M., Berg, A.C., Fei-Fei, L.: Imagenet large scale visual recognition challenge. In: International Journal of Computer Vision (IJCV), vol. 115, no. 3, pp. 1573...

    2015

  56. [64]

    In: Proceedings of the International Conference on Neural Information Processing Systems (NeurIPS) (2014) 16

    Simonyan, K., Zisserman, A.: Two-stream convolutional networks for action recognition in videos. In: Proceedings of the International Conference on Neural Information Processing Systems (NeurIPS) (2014) 16

    2014

  57. [65]

    In: arXiv preprint arXiv:1212.0402 (2012) 5, 9, 16

    Soomro, K., Zamir, A.R., Shah, M.: Ucf101: A dataset of 101 human actions classes from videos in the wild. In: arXiv preprint arXiv:1212.0402 (2012) 5, 9, 16

    2012 arXiv

  58. [66]

    In: Proceedings of the ACM Conference on Fairness, Accountability, and Transparency (2021) 1 Pre-training for Action Recognition with Automatically Generated Fractal Datasets 19

    Steed, R., Caliskan, A.: Image representations learned with unsupervised pre-training contain human-like bi- ases. In: Proceedings of the ACM Conference on Fairness, Accountability, and Transparency (2021) 1 Pre-training for Action Recognition with Automatically Generated Frac...

    2021

  59. [67]

    Thoker, F.M., Doughty, H., Bagad, P., Snoek, C.G.M.: How severe is benchmark-sensitivity in video self- supervised learning? In: Proceedings of the European Conference on Computer Vision (ECCV) (2022) 4, 11, 12

    2022

  60. [68]

    In: Proceedings of the International Conference on Neural Information Pro- cessing Systems (NeurIPS) (2023) 15

    Tian, Y., Fan, L., Isola, P., Chang, H., Krishnan, D.: Sta- blerep: Synthetic images from text-to-image models make strong visual representation learners. In: Proceedings of the International Conference on Neural Information Pro- cessing Systems (NeurIPS) (2023) 15

    2023

  61. [69]

    In: Proceedings of the In- ternational Conference on Neural Information Processing Systems (NeurIPS) (2022) 16

    Tong, Z., Song, Y., Wang, J., Wang, L.: VideoMAE: Masked autoencoders are data-efficient learners for self- supervised video pre-training. In: Proceedings of the In- ternational Conference on Neural Information Processing Systems (NeurIPS) (2022) 16

    2022

  62. [70]

    In: Proceedings of the IEEE Interna- tional Conference on Computer Vision (ICCV) (2015) 16

    Tran, D., Bourdev, L., Fergus, R., Torresani, L., Paluri, M.: Learning spatiotemporal features with 3d convolu- tional networks. In: Proceedings of the IEEE Interna- tional Conference on Computer Vision (ICCV) (2015) 16

    2015

  63. [71]

    In: Signal Pro- cessing, vol

    Vincent, L.: Morphological transformations of binary im- ages with arbitrary structuring elements. In: Signal Pro- cessing, vol. 22, no. 1, pp. 3–23 (1991) 8

    1991

  64. [72]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2021) 6

    Wang, J., Gao, Y., Li, K., Lin, Y., Ma, A.J., Cheng, H., Peng, P., Huang, F., Ji, R., Sun, X.: Removing the back- ground by adding the background: Towards background robust self-supervised video representation learning. In: Proceedings of the IEEE Conference on Computer Vision...

    2021

  65. [73]

    In: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2023) 16

    Wang, L., Huang, B., Zhao, Z., Tong, Z., He, Y., Wang, Y., Wang, Y., Qiao, Y.: Videomae v2: Scaling video masked autoencoders with dual masking. In: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2023) 16

    2023

  66. [74]

    In: Proceed- ings of the European Conference on Computer Vision (ECCV) (2016) 16

    Wang, L., Xiong, Y., Wang, Z., Qiao, Y., Lin, D., Tang, X., Van Gool, L.: Temporal segment networks: Towards good practices for deep action recognition. In: Proceed- ings of the European Conference on Computer Vision (ECCV) (2016) 16

    2016

  67. [75]

    In: arXiv preprint arXiv:1902.11097 (2019) 1

    Wilson, B., Hoffman, J., Morgenstern, J.: Predic- tive inequity in object detection. In: arXiv preprint arXiv:1902.11097 (2019) 1

    2019 arXiv

  68. [76]

    In: arXiv preprint arXiv:1905.00546 (2019) 1

    Yalniz, I.Z., J´ egou, H., Chen, K., Paluri, M., Mahajan, D.: Billion-scale semi-supervised learning for image clas- sification. In: arXiv preprint arXiv:1905.00546 (2019) 1

    2019 arXiv

  69. [77]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2022) 1, 15

    Yamada, R., Kataoka, H., Chiba, N., Domae, Y., Ogata, T.: Point cloud pre-training with natural 3d structures. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2022) 1, 15

    2022

  70. [78]

    In: Proceedings of the Conference on Empirical Methods in Natural Lan- guage Processing (EMNLP) (2017) 1

    Zhao, J., Wang, T., Yatskar, M., Ordonez, V., Chang, K.W.: Men also like shopping: Reducing gender bias am- plification using corpus-level constraints. In: Proceedings of the Conference on Empirical Methods in Natural Lan- guage Processing (EMNLP) (2017) 1

    2017

  71. [79]

    In: Proceedings of the European Conference on Computer Vision (ECCV) (2022) 15

    Zhao, K., Shen, L., Zhang, Y., Zhou, C., Wang, T., Zhang, R., Ding, S., Jia, W., Shen, W.: B´ ezierpalm: A free lunch for palmprint recognition. In: Proceedings of the European Conference on Computer Vision (ECCV) (2022) 15

    2022

  72. [80]

    In: Audio, Speech, and Language Processing, IEEE Transactions on, vol

    Zlatintsi, N., Maragos, P.: Multiscale fractal analysis of musical instrument signals with application to recogni- tion. In: Audio, Speech, and Language Processing, IEEE Transactions on, vol. 21, pp. 737–748 (2013) 16 20 Davyd Svyezhentsev et al. Hyperparameter HMDB51 UCF101 D...

    2013

Pith tools

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