Pith. sign in

REVIEW 3 major objections 6 minor 32 references

Development of Hybrid Artificial Intelligence Training on Real and Synthetic Data: Benchmark on Two Mixed Training Strategies

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Fine-tuning usually beats simple mixed training for hybrid real-and-synthetic data, but a large domain gap can reverse the result.

desk verdict A solid, controlled benchmark confirming FT usually beats SM, but the paper's own domain-gap metric contradicts its third conclusion, so that part needs major revision. read the letter →

arxiv 2506.24093 v1 pith:CHJHX4NR submitted 2025-06-30 cs.LG cs.AI

Paul Wachter , Lukas Niehaus , Julius Schöning This is my paper · ORCID
classification cs.LGcs.AI
keywords hybriddatasetsmixedtrainingfine-tuningsimpledomaingapsyntheticdataimageclassificationbenchmark
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 asks which of two ways of mixing synthetic and real images—training on both at once (simple mixed, SM) or pretraining on synthetic data and then fine-tuning on real data (FT)—better closes the reality gap in image classification. Across three datasets whose synthetic parts come from generative AI, CAD renderings, and hand-drawn sketches, three architectures, and eleven synthetic-to-real ratios, FT outperformed SM in 635 of 810 individual comparisons and in 69 of 81 averaged configurations. The one clear reversal is a CNN on the hand-drawn DomainNet sketches, where SM wins. The authors conclude that FT is usually the better strategy when the synthetic data closely mimics the real domain, and that the choice between strategies may be driven by a quantifiable domain gap. If right, this gives practitioners a rule of thumb: prefer fine-tuning for realistic synthetic data, and consider simple mixing when the gap is large.

What carries the argument

The load-bearing mechanism is a controlled benchmark design that isolates the mixing strategy from other variables. For each of three hybrid datasets, the authors create eleven equal-sized subsets spanning synthetic proportions from 0% to 100%, match class distributions across subsets, keep total dataset size constant, and train vanilla multilayer perceptron (MLP), convolutional neural network (CNN), and vision transformer (ViT) models with plain stochastic gradient descent and no augmentation or regularization. The two strategies are simple mixed (SM), which samples both domains jointly in every batch, and fine-tuned (FT), which pretrains only on synthetic data with early stopping on the real validation set and then retrains on the real part. The domain gap is operationalized as the difference between purely real and purely synthetic baseline accuracy. That operationalization, together with the controlled ratio sweep, is what lets the paper attribute performance differences to the strategy rather than to dataset size, class imbalance, or pretraining on external data.

What would settle it

Take a fourth hybrid dataset whose synthetic images are visually close to the real ones but whose synthetic-only baseline accuracy is nearly as low as the DomainNet sketches, a large measured gap with a small visual gap. If FT does not beat SM on that dataset, the proposed link between domain-gap size and strategy choice fails; alternatively, compute the correlation between the baseline gap and the FT-minus-SM advantage across many datasets and check whether a single threshold reliably separates the two regimes.

Watch

Extended reading notes

Core claim

The central discovery is that fine-tuning is generally, but not universally, superior to simple mixed training for hybrid real/synthetic image datasets, with the exception shaped by both architecture and the size of the domain gap. Using equally sized subsampled hybrid datasets with synthetic-to-real ratios from 10% to 90%, the study finds FT ahead in 69 of 81 averaged configurations, and in 635 of 810 individual comparisons. Gains from adding real data follow diminishing returns: the first 10% of real data yields the largest accuracy jump, and each further increase gives less. On the two datasets whose synthetic images were designed to resemble the real ones (CiFake from Stable Diffusion and CAD-rendered LEGO bricks), FT consistently wins; on DomainNet, where synthetic data are black-and-white sketches, a CNN trained with SM beats FT. The paper interprets this as evidence that the optimal mixing strategy interacts with the domain gap, and suggests the strategy could in principle be chosen ahead of time from a quantifiable gap measure.

Load-bearing premise

The paper's actionable conclusion rests on assuming that three datasets, one per synthetic generation mechanism, represent synthetic data generally, and that the baseline accuracy gap between purely synthetic and purely real training measures the domain gap that decides which mixing strategy wins.

Editorial extensions

If this is right

  • Practitioners mixing realistic synthetic renderings with real images should default to fine-tuning over joint training, especially when the synthetic fraction is high.
  • A small real-data addition, around 10% of the training set, captures most of the performance benefit, so budgets can start there before collecting more real samples.
  • When the synthetic domain is visually far from the real domain, as with sketches, simple mixed training can beat fine-tuning, particularly for CNNs, so the default should be re-evaluated per dataset.
  • The optimal strategy can in principle be selected ahead of training from a quantifiable domain-gap measure, turning trial-and-error into a more systematic workflow.

Reading between the lines

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

  • If the proposed domain-gap rule generalizes, any existing real/synthetic dataset pair plus a few baseline runs would be enough to check whether the FT-minus-SM gap correlates with the synthetic-versus-real baseline gap, giving a cheap way to validate the rule on new domains.
  • The kernel-level explanation for the CNN/DomainNet reversal, that early layers lock onto binary edge detectors during synthetic pretraining, predicts that lighter pretraining, domain-randomized synthetic images, or kernel reinitialization before fine-tuning should reduce or remove the SM advantage.
  • The diminishing-returns curve implies an economic rule for data collection: the marginal value of a real image is highest when the current real fraction is near zero, which could guide active learning or data-budgeting decisions.
  • Because the benchmark uses unregularized, augmentation-free networks, production models with normalization, augmentation, and modern optimizers may show a different ordering of the two strategies than the one reported here.
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. The paper presents a benchmark study comparing two mixed training strategies for hybrid real/synthetic datasets: simple mixed (SM), in which real and synthetic data are trained together, and sequential fine-tuning (FT), in which a model is pretrained on synthetic data and then fine-tuned on real data. Using three network architectures (MLP, CNN, ViT) and three hybrid datasets generated by different mechanisms (GenAI, CAD, hand drawings), the authors sample eleven synthetic-to-real proportions and report test accuracies over ten repetitions. The central empirical claim is that FT usually outperforms SM (635 of 810 individual comparisons, 69 of 81 averaged configurations), with a notable counter-example for CNNs on DomainNet, where SM performs better. The paper further claims that the optimal strategy can be selected based on a quantifiable domain gap, defined as the difference between purely real and purely synthetic baseline accuracies.

Significance. The paper contributes a controlled and fairly extensive benchmark: dataset size is held constant across ratios, class balance is enforced, baselines are external, and the full code and numerical results are made available. The repeated runs and the cross-dataset, cross-architecture design are genuine strengths, and the reported FT-over-SM advantage with a documented counter-example is a useful practical data point. However, the paper's third conclusion—that the optimal strategy can be chosen from a quantifiable domain gap—is not supported by the paper's own operationalization of that gap, which is a load-bearing issue for the advertised actionable guidance. The mechanistic explanation for the CNN/DomainNet exception is also presented without supporting evidence, and the headline comparisons lack uncertainty quantification.

major comments (3)
  1. [Sections 4–6] Section 4 states that 'the domain gap can be quantified as the difference between the purely real and purely synthetic results' (Table 2). Under that definition, DomainNet has the smallest gap, not the largest: for the CNN, Cifar-10/CiFake gap = 0.6413 − 0.1430 = 0.4983, LegoBricks gap = 0.5582 − 0.0130 = 0.5452, and DomainNet gap = 0.2870 − 0.0353 = 0.2517; the same ordering holds for the MLP and ViT. Yet Section 5 asserts 'DomainNet exhibits a larger domain gap compared to Cifar-10/CiFake and LegoBricks', and Section 6 concludes that the optimal strategy could be chosen based on a quantifiable domain gap. The stated metric contradicts the qualitative claim, so the third conclusion is unsupported by the reported data. The authors must either revise the gap measure to one that actually orders the datasets as claimed, or present the DomainNet result purely as an architecture–data interaction without linking it to gap magnitude.
  2. [Section 4] The headline comparison counts (635/810 and 69/81) are reported without confidence intervals, standard deviations, or significance tests. Several boxplots, notably Fig. 5 for CNN/DomainNet and many appendix figures, show largely overlapping distributions between SM and FT at individual ratios, so it is unclear whether the aggregate counts reflect reliable differences or are dominated by small, noisy effects. Given that each configuration was repeated ten times, paired tests or confidence intervals on the per-configuration accuracy differences should be provided to support the first conclusion that 'FT is a preferable strategy'.
  3. [Section 5] The proposed explanation for the CNN/DomainNet counter-example—that pretraining on simple black-and-white sketches biases early convolutional layers toward edge detectors, leading to ill-conditioned gradients during fine-tuning—is presented as the reason for the results, but no evidence is offered: there are no kernel visualizations, probing experiments, or ablations that would distinguish this mechanism from alternative explanations. As written, this is a plausible post-hoc hypothesis, not a substantiated finding. The authors should either clearly label the discussion as speculative or add targeted analyses to support it, especially since the paper's main contribution is its controlled experimental comparison.
minor comments (6)
  1. [Section 6] The text says 'Across 89 individual configurations FT surpassed SM in 69 (78%)', but the experimental design yields 81 averaged configurations (3 datasets × 3 architectures × 9 hybrid ratios), matching the '69 of 81' reported in Section 4. The number 89 appears to be a typo and should be corrected.
  2. [Section 5] The reference '(Fig. 3.3)' near the description of DomainNet's synthetic images appears to be an incorrect cross-reference; it should likely point to Fig. 1 or a relevant panel.
  3. [Figures 4, 5, 12–14] The figure captions label the dataset as 'domain' instead of 'DomainNet'; please use the full dataset name for clarity.
  4. [Throughout] The dataset name is written inconsistently as 'Cifake', 'CiFake', and 'Cifar-10/Cifake' in different places; standardize the capitalization and the hybrid dataset label.
  5. [Various] There are several typographical and grammatical issues, including 'Therefor' (Section 3.2), 'It's real part' (Section 3.3), 'relaying on' (Section 1), and 'sgriculture' in reference [29]. A careful copyedit is recommended.
  6. [Section 3.3] The sentence 'LEGO Bricks [2] for training classification networks was selected as the second dataset' is awkwardly phrased; consider rewording to make the subject and intent clear.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the benchmark results are measured, not derived from their own inputs.

full rationale

This is an empirical benchmark, not a derivation chain. The central claims (FT outperforms SM in 635/810 comparisons, ratio effects, baselines) are measured on held-out real test sets over ten repetitions, with purely synthetic and purely real training as external baselines. No quantity in the paper is defined in terms of the result it is meant to explain, and no fitted parameter is relabeled as a prediction. The only self-citation, [29] Wachter et al., appears in Related Work as an example of synthetic-data generation and is not load-bearing for any conclusion. For completeness, the paper does contain a consistency problem that is not circularity: Section 4 states that 'the domain gap can be quantified as the difference between the purely real and purely synthetic results' (Table 2), but Section 5 asserts 'DomainNet exhibits a larger domain gap compared to Cifar-10/CiFake and LegoBricks,' which contradicts that quantification (CNN real-minus-synthetic accuracy: Cifar-10/CiFake 0.6413-0.1430=0.4983, LegoBricks 0.5582-0.0130=0.5452, DomainNet 0.2870-0.0353=0.2517; similar ordering for MLP and ViT). Consequently, the Section 6 conclusion that the optimal strategy 'could be chosen based on a quantifiable domain gap measure' is unsupported by the paper's own metric. This is a correctness/consistency issue, not circular reasoning, because the conclusion is not assumed in constructing the measurements or in defining the strategies; the empirical comparisons themselves remain self-contained.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The central claim is an empirical comparison, so there are no fitted constants or invented entities. The axioms capture the representativeness assumptions that would be needed to generalize the benchmark findings beyond the three datasets and three architectures studied.

assumptions (3)
  • domain assumption The three selected hybrid datasets (Cifar-10/CiFake, LegoBricks, DomainNet) are representative of synthetic data generation mechanisms (GenAI, CAD, hand drawings) and their associated domain gaps.
    Section 3.3 introduces the three datasets as covering distinct synthetic generation mechanisms. The paper's third conclusion (Section 6) generalizes from these three to a broader domain-gap-strategy relationship.
  • domain assumption The difference between purely synthetic and purely real baseline test accuracy is a valid operationalization of the domain gap.
    In Section 4, the domain gap is quantified as the difference between the real and synthetic baseline results, and Section 5 uses this framing to explain why DomainNet behaves differently. This measure is not validated against any independent domain-gap metric.
  • domain assumption The training protocol (vanilla SGD, no regularization/data augmentation, fixed hyperparameters) isolates the effect of the mixing strategy.
    Section 3.6 states the protocol intentionally minimizes confounding variables. The paper assumes this allows the comparison to generalize to other networks despite later acknowledging production-grade networks may differ.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Development of Hybrid Artificial Intelligence Training on Real and Synthetic Data: Benchmark on Two Mixed Training Strategies." pith.science (2026). https://pith.science/paper/CHJHX4NR

@misc{pith2026250624093,
  author       = {Pith},
  title        = {Pith review of: Development of Hybrid Artificial Intelligence Training on Real and Synthetic Data: Benchmark on Two Mixed Training Strategies},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CHJHX4NR}},
  note         = {Machine review of arXiv:2506.24093}
}
read the original abstract

Synthetic data has emerged as a cost-effective alternative to real data for training artificial neural networks (ANN). However, the disparity between synthetic and real data results in a domain gap. That gap leads to poor performance and generalization of the trained ANN when applied to real-world scenarios. Several strategies have been developed to bridge this gap, which combine synthetic and real data, known as mixed training using hybrid datasets. While these strategies have been shown to mitigate the domain gap, a systematic evaluation of their generalizability and robustness across various tasks and architectures remains underexplored. To address this challenge, our study comprehensively analyzes two widely used mixing strategies on three prevalent architectures and three distinct hybrid datasets. From these datasets, we sample subsets with varying proportions of synthetic to real data to investigate the impact of synthetic and real components. The findings of this paper provide valuable insights into optimizing the use of synthetic data in the training process of any ANN, contributing to enhancing robustness and efficacy.

Figures

Figures reproduced from arXiv: 2506.24093 by the authors.

Figure 1
Figure 1. Real and synthetic example images of the datasets. [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. ViT test accuracy on the Cifar-10/CiFake dataset (0.0=100% real, [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Test accuracy gradients of the ViT trained on Cifar-10/CiFake datasets. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: MLP test accuracy on the DomainNet dataset (0.0=100% real, 1.0=100% [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: CNN test accuracy on the DomainNet dataset (0.0=100% real, 1.0=100% [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: MLP test accuracy on the Cifar-10/Cifake dataset (0.0=100% real, [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: CNN test accuracy on the Cifar-10/Cifake dataset (0.0=100% real, [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: ViT test accuracy on the Cifar-10/Cifake dataset (0.0=100% real, [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: MLP test accuracy on the LegoBricks dataset (0.0=100% real, 1.0=100% [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: CNN test accuracy on the LegoBricks dataset (0.0=100% real, 1.0=100% [PITH_FULL_IMAGE:figures/full_fig_p016_10.png]
Figure 11
Figure 11. Figure 11: ViT test accuracy on the LegoBricks dataset (0.0=100% real, 1.0=100% [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]
Figure 12
Figure 12. Figure 12: MLP test accuracy on the DomainNet dataset (0.0=100% real, 1.0=100% [PITH_FULL_IMAGE:figures/full_fig_p017_12.png]
Figure 13
Figure 13. Figure 13: CNN test accuracy on the DomainNet dataset (0.0=100% real, 1.0=100% [PITH_FULL_IMAGE:figures/full_fig_p018_13.png]
Figure 14
Figure 14. Figure 14: ViT test accuracy on the DomainNet dataset (0.0=100% real, 1.0=100% [PITH_FULL_IMAGE:figures/full_fig_p018_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 14 canonical work pages

  1. [1]

    IEEE Access12, 15642–15650 (2024)

    Bird, J.J., Lotfi, A.: Cifake: Image classification and explainable identification of ai- generated synthetic images. IEEE Access12, 15642–15650 (2024). https://doi. org/10.1109/access.2024.3356122

    work page arXiv 2024

  2. [2]

    https://doi.org/10.34808/RCZA-JY08

    Boiński, T., Zaraziński, S., Śledź, B.: Lego bricks for training classification network (2021). https://doi.org/10.34808/RCZA-JY08

    work page doi:10.34808/rcza-jy08 2021

  3. [3]

    Burdorf, S., Plum, K., Hasenklever, D.: Reducing the amount of real world data for object detector training with synthetic data (2022).https://doi.org/10.48550/ ARXIV.2202.00632

    work page Pith review arXiv doi:10.48550/arxiv.2202.00632 2022

  4. [4]

    https://doi.org/10.48550/ARXIV.2208.04052

    De Roovere, P., Moonen, S., Michiels, N., Wyffels, F.: Dataset of industrial metal objects (2022). https://doi.org/10.48550/ARXIV.2208.04052

    arXiv doi:10.48550/arxiv.2208.04052 2022

  5. [5]

    In: International Conference on Learning Representations (ICLR) (2021),https: //openreview.net/forum?id=YicbFdNTTy

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., Houlsby, N.: An image is worth 16x16 words: Transformers for image recognition at scale. In: International Conference on Learning Representations (ICLR) (2021),https: //openreview.net/forum?id=YicbFdNTTy

    2021

  6. [6]

    Ettedgui, S., Abu-Hussein, S., Giryes, R.: ProCST: Boosting semantic segmen- tation using progressive cyclic style-transfer (2022).https://doi.org/10.48550/ ARXIV.2204.11891

    work page Pith review arXiv doi:10.48550/arxiv.2204.11891 2022

  7. [7]

    Sensors 21(23), 7901 (Nov 2021)

    Eversberg, L., Lambrecht, J.: Generating images with physics-based rendering for an industrial object detection task: Realism versus domain randomization. Sensors 21(23), 7901 (Nov 2021). https://doi.org/10.3390/s21237901

    work page doi:10.3390/s21237901 2021

  8. [8]

    Biological Cybernetics 36(4), 193–202 (Apr 1980).https://doi.org/10.1007/bf00344251

    Fukushima, K.: Neocognitron: A self-organizing neural network model for a mech- anism of pattern recognition unaffected by shift in position. Biological Cybernetics 36(4), 193–202 (Apr 1980).https://doi.org/10.1007/bf00344251

    doi:10.1007/bf00344251 1980

Show all 32 references
  1. [9]

    Automation in Construc- tion 149, 104771 (May 2023).https://doi.org/10.1016/j.autcon.2023.104771

    Kim,J.,Kim,D.,Lee,S.,Chi,S.:Hybriddnntrainingusingbothsyntheticandreal construction images to overcome training data shortage. Automation in Construc- tion 149, 104771 (May 2023).https://doi.org/10.1016/j.autcon.2023.104771

    2023

  2. [10]

    Frontiers in Plant Science15 (Sep 2024)

    Klein, J., Waller, R., Pirk, S., Pałubicki, W., Tester, M., Michels, D.L.: Synthetic data at scale: a development model to efficiently leverage machine learning in agriculture. Frontiers in Plant Science15 (Sep 2024). https://doi.org/10.3389/ fpls.2024.1360113

    2024

  3. [11]

    Krizhevsky, A., Hinton, G.: Learning multiple layers of features from tiny im- ages. Tech. rep., University of Toronto, Toronto, Ontario (2009),https://www.cs. toronto.edu/~kriz/learning-features-2009-TR.pdf

    2009

  4. [12]

    In: 2019 19th International Conference on Advanced Robotics (ICAR)

    Lambrecht, J., Kästner, L.: Towards the usage of synthetic data for marker-less pose estimation of articulated robots in rgb images. In: 2019 19th International Conference on Advanced Robotics (ICAR). pp. 240–247. IEEE (Dec 2019).https: //doi.org/10.1109/icar46387.2019.8981600

    2019

  5. [13]

    LDraw.org: Content and licensing information (2020),https://www.ldraw.org/, accessed: 2025-04-24

    2020

  6. [14]

    Springer International Publish- ing (2021)

    Nikolenko, S.I.: Synthetic Data for Deep Learning. Springer International Publish- ing (2021). https://doi.org/10.1007/978-3-030-75178-4

    2021 doi

  7. [15]

    https://doi.org/10.48550/ARXIV.1907

    Nowruzi, F.E., Kapoor, P., Kolhatkar, D., Hassanat, F.A., Laganiere, R., Rebut, J.: How much real data do we actually need: Analyzing object detection performance using synthetic and real data (2019). https://doi.org/10.48550/ARXIV.1907. 07061

    2019 doi

  8. [16]

    In: 2019 IEEE/CVF International Conference 20 P

    Peng, X., Bai, Q., Xia, X., Huang, Z., Saenko, K., Wang, B.: Moment matching for multi-source domain adaptation. In: 2019 IEEE/CVF International Conference 20 P. Wachter et al. on Computer Vision (ICCV). pp. 1406–1415. IEEE (Oct 2019).https://doi.org/ 10.1109/iccv.2019.00149

    2019

  9. [17]

    In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

    Peng, X., Usman, B., Kaushik, N., Wang, D., Hoffman, J., Saenko, K.: Visda: A synthetic-to-real benchmark for visual domain adaptation. In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). pp. 2102–21025. IEEE (Jun 2018).https://doi.org/10...

    2018

  10. [18]

    In: 2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW)

    Poucin, F., Kraus, A., Simon, M.: Boosting instance segmentation with synthetic data: A study to overcome the limits of real world data sets. In: 2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW). pp. 945–953. IEEE (Oct 2021).https://doi.org/10.1109/i...

    2021

  11. [19]

    Rajpura, P.S., Bojinov, H., Hegde, R.S.: Object detection using deep cnns trained on synthetic images (2017).https://doi.org/10.48550/ARXIV.1706.06782

    2017 doi arXiv

  12. [20]

    In: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 10674–10685. IEEE (Jun 2022). https://doi.org/10.1109/cvpr52688.2022.01042

    2022

  13. [21]

    In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    Ros, G., Sellart, L., Materzynska, J., Vazquez, D., Lopez, A.M.: The synthia dataset: A large collection of synthetic images for semantic segmentation of urban scenes. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE (Jun 2016).https://doi.org/1...

    2016 doi

  14. [22]

    Schmidhuber, J.: Annotated history of modern ai and deep learning (2022).https: //doi.org/10.48550/ARXIV.2212.11279

    2022 doi

  15. [23]

    Applied Sciences15(1), 354 (Jan 2025)

    Staniszewski, M., Kempski, A., Marczyk, M., Socha, M., Foszner, P., Cebula, M., Labus, A., Cogiel, M., Golba, D.: Searching for the ideal recipe for preparing syn- thetic data in the multi-object detection problem. Applied Sciences15(1), 354 (Jan 2025). https://doi.org/10.3390...

    2025 doi

  16. [24]

    Springer International Publishing, 2nd edn

    Szeliski, R.: Computer Vision: Algorithms and Applications. Springer International Publishing, 2nd edn. (2022).https://doi.org/10.1007/978-3-030-34372-9

    2022 doi

  17. [25]

    In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

    Tremblay, J., Prakash, A., Acuna, D., Brophy, M., Jampani, V., Anil, C., To, T., Cameracci, E., Boochoon, S., Birchfield, S.: Training deep networks with synthetic data: Bridging the reality gap by domain randomization. In: 2018 IEEE/CVF Conference on Computer Vision and Patte...

    2018

  18. [26]

    In: 33rd British Machine Vision Confer- ence 2022, BMVC 2022, London, UK, November 21-24, 2022

    Vanherle, B., Moonen, S., Reeth, F.V., Michiels, N.: Analysis of training object detection models with synthetic data. In: 33rd British Machine Vision Confer- ence 2022, BMVC 2022, London, UK, November 21-24, 2022. BMVA Press (2022), https://bmvc2022.mpi-inf.mpg.de/0833.pdf

    2022

  19. [27]

    In: Guyon, I., Luxburg, U.V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., Garnett, R

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, L.u., Polosukhin, I.: Attention is all you need. In: Guyon, I., Luxburg, U.V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., Garnett, R. (eds.) Advances in Neural Information Processi...

    2017

  20. [28]

    In: 2022 International Conference on Smart Systems and Technologies (SST)

    Vuletic, J., Polic, M., Orsag, M.: Procedural generation of synthetic dataset for robotic applications in sweet pepper cultivation. In: 2022 International Conference on Smart Systems and Technologies (SST). pp. 309–314. IEEE (Oct 2022).https: //doi.org/10.1109/sst55530.2022.9954643

    2022

  21. [29]

    In: Informatik in der Land-, Forst-und Ernährungswirtschaft-Fokus: Biodiversität fördern durch digitale Landwirtschaft

    Wachter, P., Kruse, N., Schöning, J.: Synthetic fields, real gains: Enhancing smart sgriculture through hybrid datasets. In: Informatik in der Land-, Forst-und Ernährungswirtschaft-Fokus: Biodiversität fördern durch digitale Landwirtschaft. Development of Hybrid AI Training on...

    2024

  22. [30]

    Neurocomputing 312, 135–153 (Oct 2018).https://doi.org/10.1016/j.neucom.2018.05.083

    Wang, M., Deng, W.: Deep visual domain adaptation: A survey. Neurocomputing 312, 135–153 (Oct 2018).https://doi.org/10.1016/j.neucom.2018.05.083

    2018 doi

  23. [31]

    In: Proceedings of the 37th International Conference on Machine Learning

    Xiong, R., Yang, Y., He, D., Zheng, K., Zheng, S., Xing, C., Zhang, H., Lan, Y., Wang, L., Liu, T.Y.: On layer normalization in the transformer architecture. In: Proceedings of the 37th International Conference on Machine Learning. ICML’20, JMLR.org (2020), https://dl.acm.org/...

    2020

  24. [32]

    In: 2017 IEEE International Con- ference on Computer Vision (ICCV)

    Zhu, J.Y., Park, T., Isola, P., Efros, A.A.: Unpaired image-to-image translation using cycle-consistent adversarial networks. In: 2017 IEEE International Con- ference on Computer Vision (ICCV). pp. 2242–2251. IEEE (Oct 2017). https: //doi.org/10.1109/iccv.2017.244

    2017 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.